delivered the opinion of the Court,

After settling and dismissing Kenneth Kepple’s suit against General Tire, Inc., Kepple’s attorneys moved on their own behalf to vacate a protective order issued pursuant to Rule 166b(5)(e) of the Texas Rules of Civil Procedure,1 which restricted their disclosure of documents General produced during discovery and designated as confidential. The district court granted the motion, holding that disclosure of the documents could not be restricted unless General complied with the procedures for sealing “court records” under Rule 76a. General then moved for protection under Rule 76a, and after an evidentiary hearing, the court denied the motion, holding that General’s documents were “court records” within the meaning of Rule 76a(2)(c) and could not be sealed. The court of appeals affirmed the district court’s rulings.2 We hold that the district court erred in in-*522yoking Rule 76a’s procedures before determining whether General’s documents were “court records”, and that the court abused its discretion in determining that the documents were “court records”. Accordingly, we reverse the court of appeals’ judgment and reinstate the district court’s protective order.

I

Kyle Kepple suffered severe injuries when the Ford Bronco II in which he was riding rolled over. His father Kenneth brought suit on Kyle’s behalf against Ford Motor Company and General Tire, Inc., the manufacturer of the Bronco’s tires. Kepple claimed that a defect in one of the tires caused the tread to separate, precipitating the rollover.

On General’s motion, the district court issued an interim protective order, pursuant to Rule 166b(5)(e), that permitted General ■ to designate information produced during discovery as confidential after making “a bona fide determination that the material is, in fact, a trade secret or other confidential information, the dissemination of which would significantly damage [General’s] competitive position.” The order required Kepple to notify General if he disagreed with General’s designation of any documents. Confidential information could be disclosed only to Kep-ple’s counsel, the witnesses, the court, and the jury in this case, and to any plaintiffs’ counsel, witnesses, court, or jury in any other suit against General alleging a tire defect. Confidential information could not be disclosed to any expert witness employed by one of General’s competitors. General produced numerous documents under this order, including those containing design specifications, testing data, and “adjustment” reports detailing the frequency of returns under General’s warranty program.

Three months after issuing the protective order, the district court stated at a pretrial conference, on its own initiative, that any order limiting dissemination of the documents must comply with Rule 76a, which governs the sealing of court records. The court indicated that the protective order would be vacated and directed General, if it desired protection, to file a motion to seal the documents under Rule 76a. Although not conceding that Rule 76a applied, General filed a “motion for protective order or, in the alternative, temporary sealing order.” General asked the court to continue the terms of the interim protective order under either Rule 166b(5)(c) or Rule 76a. (General has never sought, either in the district court or on appeal, to restrict dissemination of the documents beyond the terms of the original protective order.) Before any further action was taken concerning the documents, however, the parties settled the lawsuit. The trial court signed an order dismissing the suit with prejudice, -without vacating the interim protective order.

Two months later, Kepple’s attorneys and their employees sought relief from the interim protective order, contending that General’s documents “affect the public safety under Rule 76a” and therefore “should be made public.” Over General’s opposition, the court vacated the interim protective order, but two days later issued a temporary sealing order under Rule 76a. General then filed a motion to reinstate the protections of the interim protective order under either Rule 166b(5)(c) or Rule 76a. The court scheduled a hearing on General’s motion, giving public notice as required by Rule 76a.3 Three parties intervened in the Rule 76a proceeding: Public Citizen Center for Auto Safety, Inc., Lawyers for Public Justice, and Jill Neviel, an individual.4

At the Rule 76a hearing, the court first considered evidence on whether the documents were “court records” subject to Rule 76a. After determining that the documents were court records, the court heard evidence about whether they should be sealed under the standard set forth in Rule 76a(1). Based on this evidence and an in camera review of the documents, the court determined that General had not met its burden for sealing the documents and ordered that they “be opened to the general public” with only customer names and addresses redacted.

*523General appealed.5 Only Kepple filed a brief as appellee. The court of appeals affirmed, holding that the trial court did not abuse its discretion in concluding that the documents were court records that should not be restricted from the public.6 We granted General’s application for writ of error.7

II

Rule 166b(5)(c) authorizes a court to order “that for good cause shown results of discovery be sealed or otherwise adequately protected, that its distribution be limited, or that its disclosure be restricted.” Rule 76a provides the standard and procedures for sealing court records. General first argues that Rule 76a does not apply to Rule 166b(5)(c) protective orders.

This Court promulgated Rule 76a in 1990 pursuant to legislative directive.8 The rule creates a presumption that all court records are open to the public and allows trial courts to seal court records

only upon a showing of all of the following:

(a) a specific, serious and substantial interest which clearly outweighs:

(1) this presumption of openness;

(2) any probable adverse effect that sealing will have upon the general public health or safety;

(b) no less restrictive means than sealing records will adequately and effectively protect the specific interest asserted.9

The court must hold an oral hearing, open to the public.10 The party seeking a sealing order must post public notice of the hearing at least fourteen days in advance “at the place where notices for meetings of county governmental bodies are required to be posted”.11 The notice must state the time and place of the hearing and must contain a specific description of the nature of the case and the records sought to be sealed.12 Any person has a right to intervene and be heard on the sealing question.13

Subject to certain limited exceptions, “court records” include “all documents of any nature filed in connection with any matter before any civil court”.14 Although the term generally does not include unfiled discovery, it does extend to “discovery, not filed of record, concerning matters that have a probable adverse effect upon the general public health or safety, or the administration of public office, or the operation of govern-ment_”15 This application to unfiled discovery is one of the rule’s most controversial aspects.16

Rule 166b(5) authorizes trial courts, among other things, to issue protective orders to protect trade secrets contained in discovery. The rule provides that “the court may make any order in the interest of justice necessary to protect the movant from undue burden, unnecessary expense, harassment or annoyance, op invasion of personal, constitutional, or property rights.”17 General argues that courts may exercise this authority independent of Rule 76a when the protection sought is less than a complete ban on disclosure of the documents. In support of its argument, *524General points to language in both Rule 166b(5)(c) and Rule 76a that recognizes a distinction between sealing documents and merely limiting their distribution. Rule 166b(5)(c) provides that a trial court may order that “discovery be sealed or otherwise adequately protected ” (emphasis added), such as by limiting distribution and restricting disclosure. Also, Rule 76a(1)(b) provides that a trial court, before ordering that court records be sealed, must determine that “no less restrictive means than sealing records will adequately and effectively protect the specific interest asserted.” The distinction in both rules between sealing and lesser limits on disclosure shows, General argues, that Rule 76a should not govern protective orders allowing significant disclosure of information, like the order in this case.

While General’s argument has logical appeal, it is undercut by the plain provision of Rule 166b(5)(c) that “[a]ny order under this subparagraph 5(c)' shall be made in accordance with the provisions of Rule 76a with respect to all court records subject to that rule.”18 This language leaves no leeway for interpretation. To the extent that discovery, whether filed or unfiled, is a “court record” under Rule 76a, the court must follow the stricter standards of that rule to limit its dissemination.19

However, the rules do not specifically set forth the standards governing a trial court’s threshold determination of whether particular unfiled discovery is or is not a “court record.” The district court in this case apparently concluded that it was required to apply the full range of Rule 76a procedures to this threshold determination. That is, the district court required General to post public notice, and allowed any party to intervene, with regard to the “court records” determination. Also, although the district court bifurcated the court records hearing from the hearing to determine whether the documents should be sealed, it allowed intervenors to have full access to the documents at the first stage, before it determined the documents to be court records. We conclude that the court erred in part.'

The special procedures of Rule 76a apply only to the sealing of “court records”. The language of the rule does not authorize trial courts to also apply these special procedures to the threshold determination of whether particular unfiled discovery is, indeed, a court record subject to the' rule. The rule would unnecessarily burden trial courts and litigants if it permitted a full hearing at this preliminary stage. If this were allowed, a party, merely by claiming that unfiled discovery met the standard for a court record under Rule 76a(2)(c), could trigger an elaborate, expensive process in any case where unfiled discovery has been exchanged. The opposing party would be required to post public notice, and the trial court would be required to conduct a full evidentiary hearing after at least fourteen days delay. General correctly argues that, under this interpretation, Rule 76a could easily become a tool for delay and gamesmanship. Equally likely, both courts and parties might view the rule as so cumbersome that they would make elaborate arrangements to avoid its requirements.

The rule does contemplate, however, that persons other than parties may intervene before the determination is made whether documents are “court records”. Otherwise, the parties could largely control the application of the rule without regard to the public interest the rule seeks to protect. Similarly, Rule 76a(8), which allows for immediate appeal of any order “relating to sealing or unsealing court records,” applies to the threshold “court records” determination.20 But allowing intervention before documents are determined to be “court records” does not mean that intervenors should have immediate access to the documents. ‘ On the contrary, when a party has moved to seal *525court records, or has filed a motion for protective order on discovery that may constitute a court record, trial courts should not allow intervenors access to the records over the moving party’s objection until the court determines that they are court records which cannot be sealed. Such preliminary disclosure would compromise the effectiveness of any later sealing order, possibly even mooting the controversy. Rule 76a anticipates this problem by allowing the trial court to “inspect records in camera when necessary.”21 In the present case, the district court allowed the intervenors to examine the documents General submitted in camera during the course of the Rule 76a hearing. While any harm caused to General cannot now be cured, we conclude that the trial court erred in this regard.

In summary, we hold that when a party seeks a protective order under Rule 166b(5)(c) to restrict the dissemination of unfiled discovery, and no party or intervenor contends that the discovery is a “court record,” a trial court need not conduct a hearing or render any findings on that issue. If a party or intervenor opposing a protective order claims that the discovery is a “court record,” the court must make a threshold determination on that issue. However, public notice and a Rule 76a hearing are mandated only if the court finds that the documents are court records. While a trial court is not required to determine whether unfiled discovery constitutes a court record until requested to do so by a party or intervenor, the court may raise this issue on its own motion. However, as previously discussed, a trial court may not apply the special procedures of Rule 76a (except for intervention) until it determines that the documents are court records. The district court in this case erred in applying the full range of Rule 76a procedures to the threshold “court records” determination.

Ill

Having determined that any restrictions on disclosure of General’s confidential records produced during discovery must satisfy Rule 76a, we next consider the applicable standard for review of the district court’s decision. On this issue the parties disagree, Kepple arguing that the proper standard is abuse of discretion, and General contending that the proper standard is sufficiency of the evidence. The courts of appeals have split on this issue, three courts reviewing Rule 76a decisions for abuse of discretion22 and one court reviewing such decisions for sufficient supporting evidence.23

General argues that we adopted a legal and factual sufficiency standard of review in Chandler v. Hyundai Motor Company.24 We disagree. In that case, the court of appeals dismissed an appeal brought under Rule 76a(8) relating to unfiled discovery, concluding that Rule 76a had no application to unfiled discovery. Because this conflicted with the plain language of Rule 76a(2)(c), we reversed the court of appeals’ judgment and remanded the cause to that court for further proceedings. Although we stated that the remand would permit the court of appeals to consider “any legal and factual insufficiency challenges properly raised by Chandler to the trial court’s final judgment restricting dissemination of documents”, we expressly reserved opinion on the merits of the appeal.25 The standard of review was not disputed in that case. Given that “[i]t is an abuse of discretion for a trial court to rale ... without supporting evidence”,26 our ref*526erence to “legal and factual insufficiency challenges” cannot be read to have established the proper standard of review.

We employ an abuse of discretion standard to review a trial court’s discovery rulings, including rulings on protective orders under Rule 166b.27 Although governed by stricter standards, a sealing order under Rule 76a is akin to a protective order, in that it requires the court to determine the extent to which information should be restricted from the public. Indeed, a trial court’s threshold determination of whether unfiled discovery is a “court record” under Rule 76a will often occur in the context of a motion for protection. The nature of Rule 76a comports with an abuse of discretion review standard. In determining whether court records should be sealed, a trial court is not called upon to make a factual finding per se, but rather is required to balance the public’s interest in open court proceedings against an individual litigant’s personal or proprietary interest in privacy.28 Such a decision necessarily requires the exercise of judicial discretion, and should be reviewed on that basis. Accordingly, we hold that Rule 76a decisions must be reviewed for abuse of discretion.

IV

We turn now to the documents at issue in this case. At the outset, Kepple argues that by producing its records in camera, General “filed” them with the court so that they became court records, regardless of whether they would otherwise meet the “court records” standard for unfiled discovery in Rule 76a. This argument is without merit. Rule 76a specifically allows a party to submit disputed records to the trial court for in camera review when necessary.29 Were it otherwise, trial courts could not review the documents themselves in determining how to apply Rule 76a without requiring the party with the documents to relinquish the very relief sought under the rule. We therefore hold that documents do not lose their character as unfiled discovery merely because they are submitted to the trial court for in camera inspection in the context of a Rule 76a hearing.

The documents at issue include those produced in this case and those produced in another case but made available to Kepple in the present case, referred to as the Benson documents. The parties agree that the documents produced in this case fall into seven categories: cured tire standards; specification revisions; product change proposals; developmental testing documents; tire mold drawings; tire adjustment data; and miscellaneous information. General argues that, for each of these categories, the trial court abused its discretion in determining they were court records, because they are not, under the rule, “discovery, not filed of record, concerning matters that have a probable adverse effect upon the general public health or safety, or the administration of public office, or the operation of government.”30 A trial court abuses its discretion when it acts in an unreasonable or arbitrary manner or, stated differently, when it acts without reference to any guiding rules or principles.31 “An abuse of discretion does not exist where the trial court bases its decisions on conflicting evidence.”32 For purposes of our analysis, we consider together all categories other than the tire adjustment data. We will then consider separately the tire adjustment data, and finally the Benson documents.

A

“Cured tire standards” show the design specifications for certain tires General manufactured or formerly manufactured. These documents contain such information as *527the components and materials used in the tire, dimensions and placement of those components, the manufacturing equipment to be used, and how the tire will be cured into the final product. “Specification revisions” describe the changes General made to its cured tire standards over a period of time. Similarly, “product change proposals” describe suggested changes to the design standards, with supporting test data, which standards may or may not have been adopted. “Developmental testing documents” show testing that was performed on various models of tires. “Tire mold drawings” specify the dimensions and configurations of the metal molds used to form the tires. Finally, the category “miscellaneous documents” comprises consumer inquiries, retention records showing the length of time that certain documents were retained, and product specifications received from Ford Motor Company.

Kepple’s evidence of a defect in General’s tires did not link the alleged defect to any of these documents. For example, George Edwards, a consultant who specializes in tire failure analysis, testified as an expert for Kepple. Because he had acted as an expert witness in numerous cases against General, Edwards had examined the in camera documents at issue. Edwards testified that, in his opinion, tread separations in tires produced under these design and testing documents have been responsible for numerous serious traffic accidents. He opined that a defect in the “skim stock,” which holds the steel belts together, was the root cause of the problem. Edwards reached these conclusions by analyzing actual tires involved in accidents, however, and he admitted that he could not correlate any tire defect with the information in the in camera documents. Dennis Carlson also testified as an expert for Kepple. Carlson testified generally that General changed its skim stock compound in 1987, and that any documents relating to that change would be important to the public safety. Like Edwards, however, Carlson was not able to correlate any defect in General’s tires with a specific document or set of documents. Kepple presented other evidence through depositions and affidavits supporting its claim of product defect. None of this additional evidence, however, linked any defect to specific documents.

A party cannot demonstrate that a manufacturer’s proprietary design, research, and testing records have a probable adverse effect on the public health or safety, as Rule 76a requires before documents are “court records”, merely by producing evidence of a defect in the manufacturer’s products. Rather, the party must, at a minimum, demonstrate some nexus between the alleged defect and the documents at issue. Because Kepple failed to demonstrate any such nexus, the district court abused its discretion in classifying the documents as “court records” under Rule 76a(2)(c).33

B

The remaining documents, the “tire adjustment data”, chart the frequency that General’s customers have returned certain tires under the company’s warranty program, and the reasons for the returns. Tom Lee, a former tire engineer for General, testified that General uses this data primarily as a “marketing tool” to determine customer satisfaction. He further testified that merely because a tire is returned does not mean that it is defective. According to Lee, nothing in the adjustment data would adversely affect the public health and safety.

Edwards, on the other hand, characterized adjustment data as the tire industry’s “report card” on the quality of its product. He testified that General’s adjustment data revealed an “extraordinary amount of tread separations.” Edwards further testified that tread separations can cause a vehicle to go out of control, and he testified about specific cases in which individuals had been seriously injured in accidents caused, in his opinion, by tire tread separations. Edwards thus concluded that the adjustment data did affect *528public safety. For some of the time periods reported, he claimed that he had never seen adjustment figures as high as General’s, with the exception of the Firestone 500 tire, which had 14 million recalls. However, Edwards testified that he found the adjustment data “terribly confusing”. The reason for the confusion, according to General’s expert, Lee, was that computer errors had affected the figures in one report, incorrectly combining tread separations with other reasons for adjustments and thus greatly overstating the number of adjustments attributable to tread separations. The corrected figures showed adjustments of only 0.8 percent due to separations, just over half the 1.5 percent maximum Edwards believed to be acceptable. No one challenged that the data had been properly corrected, and there is no evidence that the corrected adjustment data could adversely affect public health or safety.

Kepple also presented evidence that General failed to provide its adjustment figures to the National Highway Traffic Safety Administration during a 1993 investigation into tire safety. Although the NHTSA closed its investigation after five months without finding evidence of a defect in General’s tires, Edwards, who had reviewed the NHTSA file, testified that General did not provide that agency with its 1987 adjustment figures, even though the NHTSA had specifically requested the information. Edwards thus concluded that the NHTSA investigation was based on incomplete data, and that the 1987 adjustment records should be provided to the government as soon as possible. However, Edwards simply ignored the fact that any person may complain to the NHTSA about a tire manufacturer at any time.34 Furthermore, the NHTSA is not dependent on voluntary production of information but is empowered to use compulsory process to obtain information in its investigations.35 Restricting disclosure of the adjustment simply could not impede any federal investigation. While this case has been pending on appeal, two consumer groups have urged the NHTSA to reopen its 1993 investigation of General’s tires.36 The agency has taken no further action. “A decision by the relevant governmental agencies ... not to pursue a certain matter suggests a lack of basis for doing so.”37

In sum, there is no evidence that the tire adjustment data General produced for Kep-ple could adversely affect public health and safety. The district court therefore abused its discretion in concluding that such data were “court records” within the meaning of Rule 76a.

C

The final group of documents involved is the so-called Benson documents. During the pretrial conference, General complained to the trial court that Kepple had requested numerous documents which General had already produced to Kepple’s counsel in another lawsuit. General argued that it should not be required to produce this discovery twice to the same attorneys. The parties therefore agreed, with the court’s approval, that General would not have to produce the documents again, but that General could not object to their admissibility based on the fact that Kepple’s counsel had obtained them elsewhere. In other words, the parties agreed that the documents would be deemed to have been produced in this lawsuit.

The Benson documents were at issue at the Rule 76a hearing along with the other documents produced in thé case. The parties stipulated that the Benson documents fell into the same categories as the other discovery, and that evidence adduced at the hearing pertains equally to the Benson documents. General argues, however, that the Benson documents were produced under a separate protective order issued by the Benson court, and that the district court in this case had no jurisdiction to rescind that pro-*529teetive order. We disagree. Although the Benson documents were originally produced in a different lawsuit, General constructively produced the documents again in this lawsuit, subject to this trial court’s protective order. Under these circumstances, the district court had jurisdiction to modify or rescind its protection of the Benson documents.

The Benson documents must be treated in all respects like the other documents General produced, and thus our holdings concerning those documents apply equally to the Benson documents.

:{« * ^ # jK

For the reasons we have explained, the judgment of the court of appeals is reversed and the case is remanded to the district court for the sole purpose of reinstating the protective order issued January 14, 1994.

concurring in part and dissenting in part.

I join in the Court’s opinion except for part IV-B. I conclude that the trial court acted within its discretion in classifying the “adjustment” documents as court records subject to Rule 76a and in refusing to seal them. I accordingly dissent from that part of the Court’s judgment reinstating the protective order on the adjustment documents. I concur in the rest of the Court’s judgment.

I.

The threshold issue for the trial court was whether the adjustment documents are “court records,” defined as

discovery, not filed of record, concerning matters that have a probable adverse effect upon the general public health or safety, or the administration of public office, or the operation of government....

Tex.R. Civ. P. 76a(2)(c). As the Court recognizes, a trial court does not abuse its discretion when it bases its decision on conflicting evidence. See Davis v. Huey, 571 S.W.2d 859, 862 (Tex. 1978). That is the situation here.

George Edwards, an expert in tire failure analysis, specifically testified that the adjustment data reflected “an extraordinary amount of tread separations,” and thus affected the public’s safety:

Q: Now, I want to talk about adjustment records. What are adjustment records, Mr. Edwards?

A: Report card, I guess, on the tire industry, where they stand.

Q: Are the people that are doing the grading the customers of the tire industry?

A: Yes.

Q: Does the information that’s reflected in the adjustment records that you viewed in the Kepple case — you have viewed the adjustment records in the Kepple case, have you not?

A: I did.

Q: Is the information that you have reviewed in the adjustment records in the Kepple case, does that affect the public safety, in your view?

A: It certainly does.

Q: And why is that, sir?

A: There’s, there’s an extraordinary amount of tread separations. When I say tread separations, I have to qualify that and say tread and belt separation, and I will explain that later.

[[Image here]]

Q: All right. If you just turn randomly, or just turn to page 243 of No. 10, and you look at cumulative returns and the percent shown for those months, I don’t want you to tell me what those figures show, direct number-wise, but what I want to know is, are those figures something that could seriously affect the public safety?

A: Yes.

Q: And do you feel that it’s vitally and critically important that those figures be made public?

A: Ido.

Q: Do you think the government ought to be given those figures?

A: I certainly do.

*530The Court disregards this testimony because Tom Lee, General’s expert, disputed the accuracy of the adjustment figures on which Edwards based his opinion. On cross examination, however, Lee admitted that he had no firsthand knowledge about the underlying data. Also, the adjustment figures on which Edwards relied were those which General had supplied to Kepple during discovery. Additionally, Lee was very vague about when he discovered the alleged discrepancy.

Based on this conflicting evidence, I conclude that the trial court did not abuse its discretion in holding that General’s tire adjustment data met the standard for “court records” under Rule 76a(2)(c). See Davis, 571 S.W.2d at 863.

II.

Because I conclude that the trial court did not abuse its discretion in determining that the adjustment documents are court records, I must consider whether the trial court properly refused to seal those documents. To overcome the presumption of openness, General bears the burden of demonstrating all of the following:

(a) a specific, serious and substantial interest which clearly outweighs:

(1) this presumption of openness;

(2) any probable adverse effect that sealing will have upon the general public health or safety;

(b) no less restrictive means than sealing records will adequately and effectively protect the specific interest asserted.

Tex.R. Civ. P. 76a(1)(a), (b). As with the “court records” determination, we review the trial court’s decision not to seal the documents for an abuse of discretion.

A properly proven trade secret is an interest that trial courts should consider in determining whether to seal records under Rule 76a. Eli Lilly & Co. v. Marshall, 829 S.W.2d 157, 158 (Tex. 1992). General presented evidence that the adjustment reports contain trade secrets. Lee testified that a competitor could use General’s adjustment data to influence tire dealers to switch manufacturers. He further testified that the adjustment data shows not only the number of returns but also production figures, which another manufacturer could use for competitive purposes. Lee stated that General has always taken efforts to restrict the dissemination of its adjustment information.

Kepple presented evidence, however, that much of the adjustment data relates to a tire model that General has not manufactured since 1988. Indeed, the key adjustment figures on which the parties focused at the hearing are from 1986 and 1987. Regarding the production figures reflected in the adjustment data, Dennis Carlson, a tire failure analyst with extensive research and development experience in the tire industry, testified that he “[could not] see why it would be important” to maintain the confidentiality of production statistics of a discontinued tire model. Carlson further testified that plant production capacities are common knowledge, and are even published in trade journals.

Based on this conflicting testimony, I conclude that the trial court did not abuse its discretion in finding that General failed to meet its burden for sealing the documents under Rule 76a(l).

[[Image here]]

For the foregoing reasons, I dissent from that part of the Court’s judgment reinstating the protective order for the adjustment documents.