This is the second appeal in this litigation arising from application of a herbicide to cropland. The sticking point is determining the extent to which federal legislation, the federal insecticide fungicide and rodenticide act (FIFRA) has preempted the area. The case comes to us on further review of a court of appeals decision that left plaintiffs with only one state court remedy. We agree the plaintiffs are left with only one remedy, but disagree as to which one. We conclude the plaintiffs are free to seek recovery in our courts on their claim of negligent design and testing. We therefore affirm in part, reverse in part, and remand for determination on the merits of the one remaining claim.

In the mid-1980s, defendant American Cyanamid Company manufactured and marketed a herbicide named Scepter. Pursuant to FIFRA, American Cyanamid registered Scepter with the environmental protection agency (EPA), and the EPA approved the Scepter label submitted by American Cyan-amid. The label contained a section on “rotational crop restrictions” that told farmers how soon they could plant various crops on fields that had been treated with Scepter. Of particular significance for this ease, the label stated that corn could be planted eleven months after the last application of Scepter.

In 1987 and 1988, Clifford Ackerman, a farmer, used Scepter to control weeds in his soybean fields. He bought the herbicide from Allison-Kesley Ag Center, an independent dealer in agricultural supplies. Allison-Kesley told Ackerman that Scepter was safe *211for follow corn.1 Ackerman applied Scepter to his soybean field in 1987 and planted corn on the same fields the following spring. Although he had waited eleven months after the last application of Scepter to plant his follow corn, the 1988 corn crop did not do well. It was later learned that Scepter was causing carryover damage in some parts of the United States, apparently because conditions prevented the herbicide from degrading prior to the next crop. American Cyanamid agreed to pay Ackerman for the damage to his 1988 crop.

Ackerman applied Scepter again in 1988 and, when he again experienced difficulty with his 1989 crop, an American Cyanamid representative met with him regarding a settlement. Given the option of a preharvest or postharvest settlement, Ackerman signed a release for a preharvest settlement in the amount of $8627.92. After checking his crop though, Ackerman estimated the loss at $41,-309.40 and attempted to take the postharvest settlement option of $31,900. The release however had already been forwarded for approval and American Cyanamid ultimately determined it would only pay the preharvest settlement amount.

Ackerman filed this petition against American Cyanamid seeking damages for the carryover damage to his 1989 crop. Ackerman, in addition to a number of counts no longer at issue, sought recovery on two theories: breach of implied warranty of merchantability and negligent design and testing. The petition attacked the settlement agreement head-on, charging it was the product of fraudulent misrepresentation and breach of contract. The district court granted American Cyanamid’s motion for summary judgment on the claims no longer at issue, finding they were preempted by FIFRA. Following a bench trial, the district court dismissed the claims of breach of implied warranty and negligent design and testing, finding they were barred by the release signed by Acker-man.

Ackerman appealed. Our court of appeals affirmed in part and reversed in part. It determined the release was not a binding contract and reversed the dismissal of the breach-of-implied-warranty and negligent-design-and-testing claims. On remand the district court dismissed the two remaining claims, based on its belief it lacked subject matter jurisdiction because the claims were preempted by FIFRA.

Ackerman again appealed. Except for a theory based on oral representations we discuss later, the court of appeals dismissed the breach-of-implied-warranty claim. It also dismissed the negligent-design-and-testing claim. Both dismissals were grounded on the district court’s holding that it, under the circumstances, lacked subject matter jurisdiction to entertain the claim. The court cited Schuver v. E.I. Du Pont de Nemours & Co., 546 N.W.2d 610 (Iowa 1996), and Clubine v. American Cyanamid Co., 534 N.W.2d 385 (Iowa 1995), for authority.

We granted further review on the application of both American Cyanamid and Ackerman. Our review of a district court’s grant of a motion to dismiss is on error. Iowa R.App. P. 4; Henry v. Shober, 566 N.W.2d 190, 191 (Iowa 1997).

I. The federal preemption doctrine is grounded upon the supremacy clause of the federal constitution:

This Constitution, and the Laws of the United States which shall be made in Pursuance thereof ... shall be the supreme Law of the Land; and the Judges in every State shall be bound thereby, any Thing in the Constitution or Laws of any State to the Contrary notwithstanding.

U.S. Const, art. VI, cl. 2. Preemption may be found where congress’ intent to preempt the field is either expressly stated or implicit in congressional policies. Clubine, 534 N.W.2d at 386-87.

FIFRA is a comprehensive federal statute regulating pesticide use, sales, and labeling. Schuver, 546 N.W.2d at 612. The EPA is the administrative agency in charge of setting appropriate regulations. Wisconsin Pub. Intervenor v. Mortier, 501 U.S. 597, 601, 111 S.Ct. 2476, 2480, 115 L.Ed.2d 532, 540 (1991). Before a pesticide may be sold, it must be registered and its labeling ap*212proved by the EPA. Welchert v. American Cyanamid, Inc., 59 F.3d 69, 71 (8th Cir. 1995). The review process requires an applicant to submit a proposed label to the EPA for approval. Id. This label must address numerous concerns, including the ingredients, directions for use, and adverse effects of the product. Id.; see also 40 C.F.R. §§ 152.50, 156.10 (1997). In addition to the written material on the actual container, the term “label” also includes written, printed, or graphic material accompanying the container, to which reference is made. Clubine, 534 N.W.2d at 387; see also Welchert, 59 F.3d at 71.

FIFRA specifically sets forth the authority the states shall have concerning the labeling of pesticides:

(a) In general

A state may regulate the sale or use of any federally registered pesticide or device in the state, but only if and to the extent the regulation does not permit any sale or use prohibited by this subchapter.

(b) Uniformity

Such state shall not impose or continue in effect any requirements for labeling or packaging in addition to or different from those required under this subchapter.

7 U.S.C. § 136v (1994). It is the preemptive effect of subsection (b) that is at issue in this case.

We discussed the preemptive effect of FI-FRA on state law claims in Schuver, 546 N.W.2d at 613, and in Clubine, 534 N.W.2d at 387, in which we found that label-based common-law claims were preempted by FI-FRA. In both cases we relied heavily on Cipollone v. Liggett Group, Inc., 505 U.S. 504, 508, 112 S.Ct. 2608, 2613, 120 L.Ed.2d 407, 417 (1992), which involved preemption under the public health cigarette smoking act of 1969. The Cipollone court held the preemption clause of the act barred all state law claims predicated on the labeling required under that act, an act which, although not identical, closely parallels the language of FIFRA. Cipollone, 505 U.S. at 523-24, 112 S.Ct. at 2621, 120 L.Ed.2d at 427. We noted that both failure to warn and labeling-based claims brought as common-law causes of action against manufacturers of pesticides are preempted by § 136v.2 Schuver, 546 N.W.2d at 613; Clubine, 534 N.W.2d at 387. Other state appellate courts have also held that FIFRA preempts labeling-based common law causes of action. See, e.g., Kawamata Farms, Inc. v. United Agri Prods., 86 Hawai'i 214, 948 P.2d 1055, 1064 (1997); Hottinger v. Trugreen Corp., 665 N.E.2d 593, 598 (Ind.Ct.App. 1996); Jenkins v. Amchem Prods., Inc., 256 Kan. 602, 886 P.2d 869, 876 (1994); Hochberg v. Zoecon Corp., 421 Mass. 456, 657 N.E.2d 1263, 1265 (1995); McAlpine v. Rhone-Poulenc Ag. Co., 285 Mont. 224, 947 P.2d 474, 477 (1997); Ackles v. Luttrell, 252 Neb. 273, 561 N.W.2d 573, 577 (1997); Quest Chem. Corp. v. Elam, 898 S.W.2d 819, 821 (Tex. 1995); All-Pure Chem. Co. v. White, 127 Wash.2d 1, 896 P.2d 697, 699 (Wash. 1995).

Ackerman contends the preemptive reach of Cipollone and its progeny has been eroded by the later opinion of Medtronic, Inc. v. Lohr, 518 U.S. 470, 116 S.Ct. 2240, 135 L.Ed.2d 700 (1996). Medtronic holds that certain state common-law claims are not preempted by the federal medical device amendments of 1976. Medtronic, 518 U.S. at 492-95, 116 S.Ct. at 2255-58, 135 L.Ed.2d at 720-21. At first blush Medtronic might seem to be a retreat from the preemptive analysis employed in Cipollone, and followed in Schuver and Clubine. But five cases decided after Medtronic show that it does not alter the law regarding preemption. Four cases expressly hold that Medtronic does not change the preemptive reach of FIFRA. See *213Hawkins v. Leslie’s Poolmart, 965 F.Supp. 566, 571-72 (D.N.J. 1997); Ackles, 561 N.W.2d at 579; Lewis v. American Cyanamid Co., 294 N.J.Super. 53, 682 A.2d 724, 730-31 (N.J.Super.Ct.App.Div. 1996); Didier v. Drexel Chem. Co., 86 Wash.App. 795, 938 P.2d 364, 367 (Wash.Ct.App. 1997). In the fifth case, Grenier v. Vermont Log Buildings, Inc., 96 F.3d 559, 563 (1st Cir. 1996), the court held that this claim, “whether presented as a negligence claim or a claim for breach of implied warranty, is preempted by FIFRA.” Medtronic was cited for two propositions: (1) where congress has included an express preemption clause in the statute, courts must begin their analysis with the language of that provision; and (2) the term “requirements” presumptively includes state causes of action as well as the laws and regulations. Grenier, 96 F.3d at 563. The court never considered that Medtronic might have impliedly overruled the decisions holding that FIFRA preempts claims based on labeling. Rather the court cited a pre-Med-tronic case in support of its holding, and noted that case, unlike Medtronic, involved FIFRA itself. Id. (citing King v. E.I. Du Pont De Nemours & Co., 996 F.2d 1346 (1st Cir. 1993)).

With the Cipollone holding left intact, our task remains to identify whether Ackerman’s claims are predicated upon labeling and packaging requirements in addition to and different from those required by FI-FRA. Of course any direct challenge to the adequacy of a label or warning is preempted. We also examine whether a claim is merely another way of alleging the label or warning was inadequate. Such an indirect challenge is also preempted. See Schuver, 546 N.W.2d at 614. But under Cipollone we assume matters outside the realm of the statute’s express preemption provision are not preempted. Cipollone, 505 U.S. at 517, 112 S.Ct. at 2618, 120 L.Ed.2d at 423. There is a presumption against preemption which counsels a narrow construction of preemption provisions. Id. at 523, 112 S.Ct. at 2621, 120 L.Ed.2d at 427. In considering whether Ackerman’s claims are preempted by FI-FRA, we express no opinion on whether they are viable under state law; we assume ar-guendo that they are. Id. at 524, 112 S.Ct. at 2621, 120 L.Ed.2d at 427.

II. Count I of Ackerman’s petition states a clear claim for breach of implied warranty of merchantability under Iowa Code section 554.2314 (1997). Ackerman’s petition states in part:

5. The plaintiff had the herbicide applied to approximately 370 acres of crop ground in 1988, in Butler County, Iowa.

6. At the time of the sale of the herbicide, the manufacturer was aware of the custom of rotation of crops year to year between soybean and corn and represented that the product would not carryover and damage corn planted in a subsequent crop year.

10. In selling the Scepter product, the seller impliedly warranted that the product was merchantable.

11. The plaintiff used the herbicide in an ordinary manner and in doing so suffered loss to his corn crop in the subsequent year.

12. The product is not fit for the ordinary purposes for which it was intended in that it carried over to subsequent year’s and damaged the corn crop which was planted in a rotation after the bean crop.

Ackerman contends his claims are not label based, but a direct contention that the product cannot be applied to bean fields in Iowa without causing carryover damage to crops, and hence is not fit for the ordinary purpose for which it is intended. This, he argues, has nothing to do with what the label says or does not say. On the-contrary we think the claim has much to do with what the label says or does not say. This was our holding in Clubine where we held a similar claim of breach of implied warranty of merchantability was preempted. Clubine, 534 N.W.2d at 387. It was also our holding in Schuver. Schuver, 546 N.W.2d at 613; see also Taylor, 54 F.3d at 563; Didier, 938 P.2d at 368.

We think Ackerman’s claim does challenge the label. In essence the claim comes down to this. If the Scepter label had been different, and the waiting period between the ap*214plication of Scepter and planting of corn had been lengthened, the label would have been merchantable. In other words, American Cyanamid could have avoided liability for breach of implied warranty of merchantability by altering its label in the language regarding safe rotation of crops. We think Ackerman’s claim stands on the use of the product in accordance with label instructions and follow crop guidelines. It should be dismissed as preempted.3

III. The court of appeals determined that some portions of Ackerman’s claim were viable because they were dependent on oral representations made by American Cyanam-id which were not preempted by FIFRA. Although the court did not refer to any specific oral representation made by American Cyanamid, we assume the court was referring to a count of Ackerman’s petition which alleges American Cyanamid represented that “the product would not carryover and damage corn planted in a subsequent crop year.” There is simply no factual basis in the record for preserving a claim based on oral representations. Ackerman testified he never received any written materials, or spoke to anyone from American Cyanamid concerning the product or its label.

IV. Count II of Ackerman’s petition states a claim for negligence in marketing Scepter and for failing to warn the customer of carryover on subsequent crops. These are common-law-duty-to-warn claims and are clearly preempted by FIFRA and are not challenged on appeal. See Schuver, 546 N.W.2d at 614.

Other portions of Ackerman’s petition indicate his negligence claim is premised upon the design and production of Scepter. The petition reads in part:

2. The defendant was negligent in its production and marketing of Scepter in the following respects:

a. It failed to design a product which would not carryover and damage corn crops planted in subsequent years.

He also alleges that American Cyanamid was negligent in testing the effects of Scepter.

Cipollone held that claims for negligent testing which are based solely upon the manufacturer’s testing or research practices, not related to advertising or promotion, are not preempted by FIFRA. Cipollone, 505 U.S. at 524-25, 112 S.Ct. at 2621-22, 120 L.Ed.2d at 428; Worm v. American Cyan-amid Co., 5 F.3d 744, 747 (4th Cir. 1993). But merely to call something a design or testing claim does not automatically avoid FIFRA’s explicit preemption clause. See Gh"enier, 96 F.3d at 564. The line between a claim for mislabeling and a claim for a defective product is razor thin, and can turn on “whether one could reasonably foresee that the manufacturer, in seeking to avoid liability for the error, would choose to alter the product or the label.” Worm, 5 F.3d at 747.

American Cyanamid relies heavily on Schuver where we concluded the negligent-testing-and-marketing claim was label-based, and therefore preempted by FIFRA. Schu-ver, 546 N.W.2d at 614-15. The gist of Schu-vefs negligence claim was that the pesticide should not have been used in his county because the soil’s pH was not at the appro*215priate level. Id. at 614. Citing to Hue v. Farmboy Spray Co., 127 Wash.2d 67, 896 P.2d 682, 692-93 (Wash. 1995), we concluded this sort of caution should go on the label. So the claims of negligence were preempted because they called for additional or different label information from that required under FIFRA. Schuver, 546 N.W.2d at 615.

Many federal courts have held that when a plaintiffs negligent-design-and-testing claim does not set forth specific allegations that the product functioned improperly, or that the company was negligent in its manufacturing or testing, the claim is preempted because it is essentially predicated on the product’s labeling. See Grenier, 96 F.3d at 564-65; Taylor, 54 F.3d at 561-62; Worm, 5 F.3d at 747-48. State courts from other jurisdictions are in accord. See Barnes v. Sandoz Crop Protection Corp., 189 Ariz. 46, 938 P.2d 95, 97 (Ariz.Ct.App. 1997); McAlpine, 947 P.2d at 477-78.

The majority of courts however have held that FIFRA does not preempt design-and-testing claims. See Lyall v. Leslie’s Poolmart, 984 F.Supp. 587, 595 (E.D.Mich. 1997); Reutzel v. Spartan Chem. Co., 903 F.Supp. 1272, 1282 (N.D.Iowa 1995); Helms v. Sporicidin Int’l, 871 F.Supp. 837, 841 (E.D.N.C. 1994); Higgins v. Monsanto Co., 862 F.Supp. 751, 757-59 (N.D.N.Y. 1994); Jillson v. Vermont Log Bldgs., Inc., 857 F.Supp. 985, 991-92 (D.Mass. 1994); Bingham v. Terminix Int’l Co., 850 F.Supp. 516, 521-22 (S.D.Miss. 1994); Wright v. How Chem., 845 F.Supp. 503, 509-11 (M.D.Tenn. 1993); Cantley v. Lorillard Tobacco Co., 681 So.2d 1057, 1061 (Ala. 1996); Romah v. Hygienic Sanitation Co., 705 A.2d 841, 854-56 (Pa.Super. 1997); Fide v. E.I. DuPont de Nemours & Co., 542 N.W.2d 769, 772 (S.D. 1996); All-Pure Chem. Co. v. White, 127 Wash.2d 1, 896 P.2d 697, 702 (Wash. 1995).

There are many factors which militate in favor of finding that Ackerman’s negligent-design-and-testing claim is not preempted by FIFRA. Ackerman’s claim is distinguishable from Schuver in that Acker-man also alleges a design defect in the production of Scepter. Ackerman’s claim does not call for additional or different information on the label from that required by FI-FRA. Ackerman also directly assails the design and testing of Scepter. This is illustrated in two ways. First, Ackerman’s interrogatory answers indicate the claim is only predicated on the product itself and not the labeling:

Interrogatory No. 12: Describe and explain in detail the alleged defect in the Scepter alleged in your petition at law.

Answer:

The defect in Scepter was that it carried over to the following crop year and negatively affected the corn yield.

Interrogatory No. 14: Explain in detail your allegations in your petition at law that defendant American Cyanamid was negligent and the factual basis therefore.

Answer:

American Cyanamid should have discovered through testing that there was a carryover problem.

Second, because this case has already been tried before the court, we have the benefit of trial testimony. Ackerman’s expert testified at length concerning the testing and design problems of Scepter. He pointed to numerous studies which indicated American Cyan-amid knew Scepter caused carryover damage and was not adequately degradable in certain weather conditions. Yet American Cyanam-id rushed Scepter onto the market so farmers would purchase its product rather than that of their competitors. In essence Acker-man does not simply allege that adequate testing would have caused American Cyan-amid to alter the Scepter label, but directly asserts that adequate testing and proper design would have caused American Cyanamid to alter the product itself.

By recognizing the negligent-design-and-testing claim, we are not requiring information on the Scepter label “different from” or “in addition to” the information FIFRA requires. We are simply recognizing an affirmative duty on manufacturers of potentially dangerous chemicals to guard against design or manufacturing defects in their chemicals. This does not interfere with FIFRA’s purpose of requiring uniform national standards of labeling.

*216Further, by recognizing claims for negligence such as Ackerman’s here, we are only requiring that pesticides such as Scepter be safely designed. This falls neatly within the FIFRA “savings” clause:

A state may regulate the sale or use of any federally registered pesticide or device in the state, but only if and to the extent the regulation does not permit any sale or use prohibited by this subchapter.

7 U.S.C. § 136v(a). Imposition of a common-law duty of care in the design and testing of harmful chemical products does not permit any sale or use prohibited by FIFRA, nor does it in any way frustrate the will of congress. FIFRA does not preempt Acker-man’s claim for negligent design and testing.

V. Dismissal of Ackerman’s other claims are not challenged or involved in this appeal. The only surviving one is his claim for negligent design and testing. Dismissal of that claim is reversed and the case is remanded for a determination of its merits on the basis of the record made during the bench trial. We are disinclined to determine, as we did in Clubine, that Ackerman should lose his claim on the merits because testing is so closely superintended by the EPA. See Clubine, 534 N.W.2d at 387-88. Especially under the facts here, where the claims rest on challenges both to design and to testing, a summary resolution by an appellate court is inappropriate. Most, though possibly not all, negligent-design-and-testing claims should turn on their merits. This process would be thwarted if, after holding such a claim is not preempted, we nevertheless dismiss as a matter of law because of EPA supervision. This was not our intent in our dispositional holding in Clubine, and any intimation to the contrary is overruled. In order to justify dismissal at the appellate level, future eases will have to reveal the extent of EPA supervision and disclose how the agency in effect supplanted the manufacturer in matters of design and testing.

DECISION OF COURT OF APPEALS VACATED; DISTRICT COURT JUDGMENT AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.

(concurring in part and dissenting in part).

I concur in the majority’s opinion with the exception of division IV, from which I respectfully dissent. My review of the case law from other jurisdictions, of our own prior cases interpreting FIFRA preemption, and of the evidence introduced at the trial of this case convinces me that the negligent design and testing claims asserted by Ackerman are disguised label-based claims that are preempted by FIFRA.

I. Applicable Legal Principles Regarding FIFRA Preemption.

Before I address the basis for my disagreement with the majority opinion, it is helpful to establish our common ground. I agree with the majority that indirect as well as direct challenges to the adequacy of the label are preempted by FIFRA. Thus, as the majority observes, simply calling a claim “a design or testing claim does not automatically avoid FIFRA’s explicit preemption clause.” Consequently, as the majority also states, our task is to determine whether a particular claim “is merely another way of alleging the label or warning was inadequate.” I even agree with the test adopted by the majority for making this determination: “whether one could reasonably foresee that the manufacturer, in seeking to avoid liability for the error, would choose to alter the product or the label.” Worm v. American Cyanamid Co., 5 F.3d 744, 747-48 (4th Cir. 1993). The overriding deficiency I see in the majority opinion is a failure to apply these principles to the evidence introduced at the trial of this case. Before discussing this issue, however, I must express my disagreement with a statement made by the majority and the erroneous implication flowing from that statement.

After setting forth the principle that design and testing claims “predicated on the product’s labeling” are preempted, the majority opinion states, “The majority of courts *217hoivever have held that FIFRA does not preempt design-and-testing claims.” (Emphasis added.) This statement implies that the court’s prior discussion of the law with respect to the preemption of disguised labeling claims represents a minority position — an implication that is not true. Courts that have considered the question of whether a claim is actually based on a problem in the design or testing of a product or, alternatively, is a disguised labeling claim have concluded that so-called design and testing claims that are actually label-based are preempted. Again, there does not appear to be any disagreement among the courts considering this issue. Moreover, the only test suggested thus far for separating true design and testing claims from disguised labeling claims is the rule stated in Worm: whether the manufacturer, in seeking to avoid liability, would alter the label or the product.

A review of the eases cited by the majority in support of its “majority-of-the-courts” statement makes it clear that there is no disagreement among courts on the basic principles governing FIFRA preemption of design and testing claims. All but one of the cases cited by the majority arose in the context of a summary judgment motion directed, at least in part, to a design or testing claim.1 Several of these eases do not even discuss disguised labeling claims. See Reutzel v. Spartan Chem. Co., 903 F.Supp. 1272 (N.D.Iowa 1995); Jillson v. Vermont Log Bldgs., Inc., 857 F.Supp. 985 (D.Mass. 1994); Wright v. Dow Chem. U.S.A., 845 F.Supp. 503 (M.D.Tenn. 1993); Cantley v. Lorillard Tobacco Co., 681 So.2d 1057 (Ala. 1996); Eide v. E.I. Du Pont de Nemours & Co., 542 N.W.2d 769 (S.D. 1996). The courts in the cases just listed merely rely on the general rule that design and testing claims are not preempted by FIFRA. They do not discuss the possibility that the claims before them are disguised labeling claims. Therefore, these cases simply do not address the issues before us in this case — are disguised labeling claims preempted and how does one determine whether a claim is a disguised labeling claim or is truly based on design or testing deficiencies? Because disguised labeling claims are not even discussed in these cases, they surely do not stand for the proposition that disguised labeling claims are not preempted, nor do they support a conclusion that there is a split of opinion on this question.

Two similar cases deserve separate consideration. The cases of Higgins v. Monsanto Co., 862 F.Supp. 751, 757 (N.D.N.Y. 1994), and Bingham v. Terminix International Co., 850 F.Supp. 516, 517 (S.D.Miss. 1994), also arose in the context of summary judgment motions directed, in part, to design and testing claims. But unlike the foregoing cases, the courts in Higgins and Bingham addressed an argument made by the defendants that suggested the negligent testing claims should be preempted because they “implicat[ed] labeling issues.” Higgins, 862 F.Supp. at 758; accord Bingham, 850 F.Supp. at 521. The defendants’ arguments were based on the rationale that “additional testing might disclose the need for further warnings.” Higgins, 862 F.Supp. at 758; accord Bingham, 850 F.Supp. at 521. The courts were unwilling to extend preemption to such a situation and rightly so. Higgins, 862 F.Supp. at 758; Bingham, 850 F.Supp. at 521. Where information that might generate additional warnings is unknown to the defendant because of inadequate testing, the failure is one of testing, not labeling.2 Thus, *218Higgins and Bingham do not represent a position contrary to the rule that disguised labeling claims, even though designated design or testing claims, are preempted.

The other cases cited by the majority lend even less credence to the idea that there is some confusion as to how design and testing claims should be addressed for purposes of FIFRA preemption. In Lyall v. Leslie’s Poolmart, 984 F.Supp. 587 (E.D.Mich. 1997), the plaintiff alleged that chlorine tablets he used for his swimming pool and the container in which they were packaged were negligently designed. 984 F.Supp. at 590. The plaintiff was severely injured when he attempted to open a container of the tablets and the lid blew off due to the buildup of toxic gas in the closed container. Id. In resisting the defendants’ motion for summary judgment, the plaintiff relied on testimony from his expert that “the chlorine product itself was defective, based on the alleged failure to test and analyze the product to determine the circumstances under which gas would be re-leased_” Id. at 595.

The court measured this proof against the test set forth in Worm for distinguishing a mislabeling claim from a defective product claim. Noting that the line between such claims “may sometimes be blurry,” the court focused on whether the defendant “could avoid liability by altering the label or the product.” Id. at 596 (citing Worm, 5 F.3d at 748). The court stated: “Where the product itself, not the label accompanying the product, would have to be changed to avoid liability, FIFRA preemption does not apply.” Id. The court concluded that was the situation before it and, therefore, the plaintiffs negligent design claims were not label-based and were not preempted. Id. at 597.

The Worm test or factor was also cited in Helms v. Sporicidin International, 871 F.Supp. 837, 843 (E.D.N.C. 1994). In Helms, the plaintiff contended that a sterilizing solution she used in the workplace was defective in its design because the product emitted toxic gases. 871 F.Supp. at 840-41. The court applied the Worm test to determine whether the plaintiffs defective product claim was preempted. Id. at 843. Focusing on “an examination of the relevant facts,” the court noted the defendant had produced all of the documents supplied to the EPA prior to registration, yet “[n]one of these documents reveal[ed] any tests to determine whether [the product] was capable of off-gassing toxic levels of [toxic chemicals].” Id. The court also highlighted the testimony of the plaintiffs expert that the plaintiff had been exposed to unsafe levels of the toxic chemical emitted by the product, and that her exposure would have been substantially less if the product had been sold in a diluted form. Id. (The testimony showed that the product had to be diluted prior to use and could have been shipped in a diluted form. Id. at 841.) In holding that the plaintiffs claim was not preempted as a labeling claim, the court paraphrased Worm, stating, “To avoid liability based on [the plaintiffs] evidence, ‘one could reasonably foresee that [the defendant] ... would choose to alter the product ...,’ instead of the labeling.” Id. at 843 (quoting Worm, 5 F.3d at 747-48). As I will discuss later, the factual patterns of Lyall and Helms are easily distinguishable from the facts of the case before us.

The final decision cited as the “majority” view is Romah v. Hygienic Sanitation Co., 705 A.2d 841 (Pa.Super.Ct. 1997). In considering the defendant’s summary judgment motion based on FIFRA preemption, the Pennsylvania Superior Court set forth the governing principles:

FIFRA pre-empts “any state common law cause of action that rests on an alleged failure to warn or convey information about a product through its label.” But claims unrelated to labelling, such as those founded on the testing, manufacture or formulation of the pesticide, are not preempted. ...

For circumstances where it is not clear whether the claim is pre-empted as related to labelling, the courts have devised a simple test. The primary means for determining the line between a pre-empted claim and a permissible claim is “whether one could reasonably foresee that the manufacturer, in seeking to avoid liability for the error, would choose to alter the product or the label.”

*219Romah, 705 A.2d at 850-51 (citations omitted). Based on these principles of law, the court characterized the decision before it as whether the plaintiffs’ “claim that [the defendant] negligently manufactured and tested [the product] is really a disguised failure to warn claim.” Id. at 853. Without discussing the record in any detail, the court concluded that the plaintiffs did not “simply allege that adequate testing would have caused [the defendant] to alter the [product] label, but, actually, allege[d] that adequate testing and proper design would have caused [the defendant] to alter the product itself so it would be safe.”3 Id. at 855. This claim, the court concluded, was not preempted. Id.

In summary, the cases cited by the majority are entirely consistent with the following propositions: (1) negligent design and testing claims are not preempted by FIFRA unless they are disguised labeling claims, and (2) disguised labeling claims are identified by determining whether “the product itself, not the label accompanying the product, would have to be changed to avoid liability_” Lyall, 984 F.Supp. at 596 (citing Worm, 5 F.3d at 747).

Our prior cases are also consistent with these legal principles. See Schuver v. E.I. Du Pont de Nemours & Co., 546 N.W.2d 610 (Iowa 1996); Clubine v. American Cyanamid Co., 534 N.W.2d 385 (Iowa 1995). In Clubine, our court adopted and applied the rule that “label-based claims” are preempted by FIFRA. 534 N.W.2d at 387. We affirmed the trial court’s dismissal of the plaintiffs failure to warn claims on preemption grounds and the dismissal of the plaintiffs negligent testing claim on the merits for failure of proof. Id. at 387-88. In Schuver, we discussed for the first time the possibility that a claim, which did not expressly criticize the label, was nevertheless label-based. 546 N.W.2d at 615. In that case, the plaintiffs’ negligence claim included allegations that the product — an herbicide applied to the plaintiffs’ cropland — was not adequately tested. Id. at 613. In addition, the plaintiffs’ strict liability claim alleged that the product “was defective and unreasonably dangerous by reason of its design, testing, inspection, manufacture and failure of warnings.” Id. at 615. We affirmed a summary judgment ruling that dismissed both claims on the basis of preemption. Id. at 616. We concluded all of the plaintiffs’ claims were label-based. Id. at 615.

Thus, the cases cited by the majority, as well as our own cases, support the conclusion that where a plaintiff has made a claim of negligent testing or defective design the court must examine the record to determine the true nature of the claim being asserted. It is in this duty that I think the majority has fallen short. I start my discussion with a review of the precise claims made in this case to which the preemption law must be applied.

II. Procedural Background and Nature of Claims.

The procedural posture of this case cannot be overemphasized. The district court decision from which this appeal is taken was rendered after a full trial on the merits. Thus, Ackerman has had every opportunity to introduce evidence to support his design and testing claims. Our task at this point, therefore, is to examine the evidence introduced at trial to determine whether the plaintiff proved a design or testing claim unrelated to the label, or whether, as the trial court found, his proof merely established “that the defendant should have *220warned against using the product” under certain conditions.

Because this case was tried to the court, we do not have the benefit of jury instructions that clarify the precise claims considered by the trial court. Therefore, we must examine the state of the pleadings at the time of trial. The negligence count of Acker-man’s petition asserts that American Cyan-amid “was negligent in its production and marketing of Scepter [because] [i]t failed to design a product which would not carryover and damage corn crops planted in subsequent years.” American Cyanamid submitted an interrogatory to Ackerman in an apparent attempt to ascertain the exact nature of this claim:

Interrogatory No. 14: Explain in detail your allegations in your petition at law that defendant American Cyanamid was negligent and the factual basis therefore.

Answer:

American Cyanamid should have discovered through testing that there was a carryover problem.

Thus, the design and testing claims at issue in this case are based on American Cyanam-id’s (1) design of a product that had a carryover problem, and (2) alleged negligent testing of its product.

III. Negligent Design Claim.

A. Negligent design law. A claim of design negligence is shown by proof that the product “was unreasonably dangerous because of defendant’s failure to use reasonable care in its design.”4 Chown v. USM Corp., 297 N.W.2d 218, 220 (Iowa 1980). We have held that the “unreasonably dangerous” element of a negligent design casé is the same as the “unreasonably dangerous” element of a strict liability claim. See id.; accord Hillrichs v. Avco Corp., 478 N.W.2d 70, 75-76 & n. 2 (Iowa 1991) (noting in that case “the strict liability claim depend[s] on virtually the same elements of proof as are required to establish the negligence claim” and making the further observation that “a growing number of courts and commentators have found that, in cases in which the plaintiffs injury is caused' by an alleged defect in the design of a product, there is' no practical difference between theories of negligence and strict liability”).

In deciding whether the evidence supports a finding that a product was “unreasonably dangerous,” we apply the principles set forth in section 402A of the Restatement (Second) of Torts. See Fell v. Kewanee Farm Equip. Co., 457 N.W.2d 911, 916 (Iowa 1990) (quoting definition of “unreasonably dangerous” from Restatement (Second) of Torts § 402A cmt. i, at 352 (1965)). Comment j to section 402A explains the interrelationship between the unreasonably dangerous element of a defective design claim and the existence of directions or warnings concerning use of the product:

In order to prevent the product from being unreasonably dangerous, the seller may be required to give directions or warning, on the container, as to its use_ ■

Where warning is given, the seller may reasonably assume that it will be read and heeded; and a product bearing such a warning, which is safe for use if it is followed, is not in defective condition, nor is it unreasonably dangerous.

Restatement (Second) of Torts § 402A cmt. j, at 353 (1965) (emphasis added).

B. Evidence of negligent design. With respect to the design claim, Ackerman’s expert "witness, Robert Ascheman, testified that an herbicide should “dissipate[] from the environment in a manner that doesn’t injure subsequent crops.... ” He testified that Scepter did not do this in certain areas of the country because the herbicide would not degrade unless there was adequate moisture and warm temperatures — factors not always present during the short growing season of the upper Midwest.

Ascheman’s testimony with respect to the standard for designing a soybean herbicide is *221also important. When Ascheman was specifically asked “what the general standards or procedures are as they relate to an ordinary prudent manufacturer in designing and testing soybean herbicides,” he answered: “And then in terms of carryover, it requires again a review of all the major crops that are likely to follow that crop in the major use areas and to be certain that the labeling is in conformity loith those determinations.” (Emphasis added.) In fact, Ascheman testified that the Scepter label was subsequently changed in November 1988 to eliminate any preplant incorporated use in “the northern area ... including Iowa” and to lower the rate for postemergence application in the same area. He also stated Ackerman’s use of the product would not have fallen within these limitations had they been on the label when Ackerman used the product.

C. Application of law to evidence. Ascheman’s testimony clearly establishes that Ackerman’s design negligence claim is predicated on the label. To impose liability on American Cyanamid under a theory of negligent design, Ackerman had to prove the product was unreasonably dangerous. See Chown, 297 N.W.2d at 220. Assuming he did so, we must then ask, for purposes of FIFRA preemption, whether American Cyanamid would have to change the product or the label to avoid this liability. See Worm, 5 F.3d at 747-48. Under the applicable products liability law, it is clear that if American Cyanamid had placed directions or warnings on its label that, if followed, would have made the product safe for use, the product would not be unreasonably dangerous and liability would be avoided.5 Ackerman’s own evidence supports the conclusion that this is exactly what a reasonable manufacturer would do: his expert testified that American Cyanamid altered its label to prohibit uses such as Ackerman’s. Moreover, there was no evidence introduced at trial that Scepter was so defectively designed that it could not safely be used in other parts of the United States where the moisture and temperature conditions are different. Thus, Ackerman’s negligent design claim is label-based and preempted by FIFRA. See Grenier v. Vermont Log Bldgs., Inc., 96 F.3d 559, 561, 564-65 (1st Cir. 1996) (evidence showed wood preservative was unfit for residential use; design defect claim, based on allegation that “it was foreseeable that [the product] would be used on residences and it was unfit for this use,” was preempted because it was “no more than an attack on the failure to warn against residential use”).

IV. Negligent Testing.

A. Negligent testing law. Our case of West v. Broderick & Bascom Rope Co., 197 N.W.2d 202 (Iowa 1972), contains this court’s most comprehensive discussion of a negligent testing claim in the context of products liability. In West, the plaintiff, an ironworker, was injured when a wire sling manufactured by the defendant broke. 197 N.W.2d at 208. He filed suit, alleging a negligent failure to warn and negligent testing of the product. Id. Both claims were submitted to the jury, which returned a substantial verdict in the plaintiffs favor. Id. On appeal, the defendant claimed the trial court erred in submitting the negligence claims to the jury. Id. We found no error in the submission of the failure-to-warn claim, but we concluded the trial court did err in submitting the negligent testing claim. Id. at 212, 215. Our analysis of the plaintiffs negligent testing claim and the evidence in support of it is highly rele*222vant here because we concluded in West that the plaintiffs testing claim was really a failure-to-warn claim.

In considering the plaintiffs testing claim, we made the following observation:

Failure to test arises in two main factual situations: where adequate tests for defects are not conducted in production and an article containing a defect is marketed, and where the particular article is not defective but the properties of the product in general are not adequately tested before it is released to the public.

Id. at 212. The claim involved in West fell in the latter category, more thoroughly described by our court as follows:

As to the second situation — testing to ascertain the properties of a product — the commonest case is one in which a new product such as a compound is launched on the market without adequate testing to ascertain harmful effects when the product is used in various ways.

Id. at 213. This court discussed two ways in which the plaintiffs claim could arguably fit within this definition: (1) a failure by the defendant to ascertain the load capacity of the wire sling; and (2) a failure by the defendant to determine any reduced capacity of the sling after use or misuse. Id.

As to the first scenario, we noted that the evidence revealed the defendant had conducted tests and experiments that showed the safe capacities of its wire slings. Id. We concluded, therefore, that the defendant’s “neglect, if there was neglect, was not in failing to ascertain information on capacities by testing, but in failing adequately to bring that information home to users.” Id. Consequently, submission of the negligent testing claim could not be supported on this basis.

Turning to the second scenario, we concluded that any failure to test for reduced capacities in the defendant’s wire slings was immaterial. Id. We stated:

Failure to test is not material under the circumstances of this case. For testing to be material, substantial evidence must be introduced of a defect in the article — in this case, a decrease in the strength of the sling. But no such evidence was introduced here. On the contrary, the evidence shows the sling possessed substantially the ultimate tensile strength of a new one. The sling [broke] because of a load beyond such tensile strength.

Moreover, the evidence shows the iron-workers would not have used this sling had they been warned of its rated capacity. Thus the relevant negligence on [the defendant’s] part, if there was negligence, was failure to give such warning, not failure to test.

Id. at 213-14. We concluded that the specification of negligence based on failure to test should not have been submitted to the jury and, therefore, reversal was required. Id. at 214-15.

B. Evidence of negligent testing. The law set forth in West requires that we examine the evidence introduced at the trial of the case before us to determine whether Acker-man actually submitted proof of a material failure to test on the part of American Cyan-amid, or whether he merely proved a failure to warn, as did the plaintiff in West. The answer to this question, under the preemption principles established above, will determine whether the plaintiff has proved a true testing claim, or a disguised labeling claim.

Ackerman’s expert witness, Robert Asche-man, identified several deficiencies in the testing methodology employed by American Cyanamid. Notwithstanding any problems in these tests, Ascheman testified in detail that many tests and experiments conducted by various groups, including American Cyan-amid, showed an unacceptable carryover problem when Scepter was used in the northern Midwest. More importantly, Ascheman testified unequivocally that American Cyan-amid was aware of the results of these tests and experiments prior to marketing Scepter in Iowa and had actually summarized them in a document distributed to university and extension people. He expressly testified that American Cyanamid’s own testing data showed a carryover problem. He also stated it was this very problem of unacceptable carryover that Ackerman experienced with his follow corn crop.

*223C. Application of law to facts. I submit that the evidence introduced by the plaintiff in support of his negligent testing claim demonstrates a disguised labeling claim. It is clear from Ascheman’s testimony that Acker-man’s testing claim is of the second variety identified in West, “the properties of the product in general are not adequately tested before it is released to the public,” 197 N.W.2d at 212, because there is no evidence that the particular containers of Scepter applied to Ackerman’s land suffered from a production defect. Thus, we are dealing with the same type of negligent testing claim we considered in West. Coincidentally, Acker-man’s negligence claim suffers from the same problem as did the negligent testing claim in West. American Cyanamid, like the defendant in West, conducted tests that showed the very information material to the plaintiffs harm — here, the carryover problem with Scepter in the upper Midwest.6 Thus, American Cyanamid’s “neglect, if there was neglect, was not in failing to ascertain information on [carryover], but in failing adequately to bring that information home to users.” Id. at 213. This claim is label-based and, therefore, preempted.

This conclusion is entirely consistent with the general principles of FIFRA preemption discussed earlier. Under these principles, we should ask whether American Cyanamid could have avoided liability for the claim by merely changing the label or whether it would have had to change the product. The answer to this question is clear. As Acker-man’s own expert testified, the carryover problem was limited to certain areas of the country; there was no evidence that the herbicide could not be used anywhere. As one would reasonably expect, American Cyanam-id simply changed the label to eliminate uses such as Ackerman’s that triggered the carryover problem. And, as Ascheman testified, had these use limitations been on the label when Ackerman applied the product to his crops, his use would have been “off-label or illegal.” Thus, American Cyanamid could have avoided liability for Ackerman’s carryover damage had the label contained the limitations on use subsequently added. Ack-erman even acknowledges in his written argument concerning his design and testing claims that “[i]n essence, Scepter was a mislabeled product as it related to use by pre-plant incorporation application in Iowa.”

Based on the record before us, it is reasonably foreseeable that American Cyanamid, in seeking to avoid liability for the carryover effects of Scepter in the upper Midwest, would alter the label, as it did, and not the product. Therefore, Ackerman’s negligent testing claim is preempted. Cf. Barnes v. Sandoz Crop Protection Corp., 189 Ariz. 46, 938 P.2d 95, 96, 98 (Ariz.Ct.App. 1997) (court observed plaintiff introduced no evidence that a cotton herbicide “should not be used on cotton grown anywhere"; court held breach of warranty claim based on allegation that cotton herbicide “was not fit for use in Arizona" was a disguised labeling claim; court also held negligent design and testing claims were preempted (Emphasis added)).

As a final matter, I must comment on the majority’s failure to follow our decision in Schuver, a decision the trial court found controlling. In my opinion, the case before us cannot be distinguished from the facts presented to us in Schuver where we held negligent testing and marketing claims were preempted. 546 N.W.2d at 615-16. The evidence in Schuver showed that the herbicide involved there, “Preview,” caused carryover damage when used on soils with a pH higher than 6.8, a fact known to the manufacturer. Id. at 614. Schuver’s land was located in O’Brien County, Iowa, a location that tended to have a pH greater than 6.8. Id. We held that Schuver’s claim that the manufacturer was negligent “in failing to test Preview in O’Brien County soil types before releasing Preview for sale in O’Brien County” and in marketing Preview in O’Brien County was preempted by FIFRA. Id. at *224613-14. We said, We think this is merely another way of arguing that Du Pont’s labels should have warned against using Preview in O’Brien County.” Id. at 614.

Our holding in Schuver dictates the same result here.7 Ackerman is complaining about a carryover problem that his own expert testified American Cyanamid knew about when it marketed Scepter in Iowa. Because this problem surfaced only in certain areas of the country, it could easily be handled in the label, as later changes in the label demonstrate. As I have shown above, any complaints about testing, designing, or marketing Scepter are really just “another way of arguing that [American Cyanamid’sJ labels should have warned against using [Scepter] in [Iowa].” Id. Thus, Ackerman’s claims are label-based, just like Schuver’s. Consequently, the trial court correctly held that they were preempted under FIFRA. I would affirm the trial court’s dismissal of Ackerman’s claims.

CARTER and LAVORATO, JJ., join this dissent.