Miguel Perez-Montes, a citizen of Mexico, entered the United States in 1989 as a lawful permanent resident. In 2001 he joined the Army and later served two tours in Afghanistan. He received a general discharge under honorable conditions. During all the years he could do so, he did not apply for citizenship. His eligibility ended in 2010, when he was convicted of a cocaine offense. That conviction led to removal proceedings and made Perez-Montes ineligible for most forms of relief.

Aliens convicted of aggravated felonies remain eligible for deferral of removal under the Convention Against Torture. Perez-Montes contended that he was at risk of being tortured or killed in Mexico because his military training would lead drug gangs to recruit him. If he refused to cooperate—he says that he would not cooperate and that the police would fail to protect him—he would be harmed. He also asserted that the Mexican government tortures its citizens who return after serving in the U.S. military. An immigration judge concluded that Perez-Montes had not established a substantial risk that he would be targeted by gangs or harmed if he refused to help them. The IJ added that, if approached by gangs, Perez-Montes could move to parts of Mexico where they don’t *850 operate, and that there is no evidence that Mexican officials mistreat former soldiers. The Board of Immigration Appeals agreed with the IJ and left the removal order,in place.

Perez-Montes does not. contend that the administrative decision, is unsupported by substantial evidence. Instead he makes a purely legal argument: that both the IJ and the BIA misunderstood the burden that an alien faces when seeking relief under the Convention. Regulations require an alien to show that torture is “more likely than not”. 8 C.F.R. §f 1208.16(b)(1)(iii), (b)(2), (c)(2), (c)(4), 1208.17(a). Perez-Montes contends that the Board and the IJ erred by asking, instead, whether he faced a “substantial risk” of torture in Mexico. That differs from the regulatory standard, he asserts, and saddled him with a greater burden.

The IJ and BIA did not pluck this phrase out of the air. It comes from Rodriguez-Molinero v. Lynch, 808 F.3d 1134 , 1136 (7th Cir. 2015), which discussed the fact that the regulatory phrase sometimes has been seen as requiring statistical proof that quantifies the precise risk an alien faces. Does the risk exceed 50%? What if there is a 20% risk of death and a 40% risk of bodily injury? Is that “more likely than not” when neither risk exceeds 50%? Does a 20% risk of death exceed a 60% risk of losing a limb? Similar questions are easy to spin out. The panel in Rodríguez-Mo-linero stated that a statistical requirement cannot be taken seriously and that the best an agency or court can do is look for substantial risk.

“More likely than not” is the standard burden in civil litigation and does not impose a statistical or quantitative requirement in a tort or contract suit any more than in a removal proceeding. Our opinion in Rodriguez-Molinero did not suggest that “substantial risk” means something more than the “more likely than not” standard. It was designed, rather, as a non-quantitative restatement of that standard. If there is any gap between the two, it is in the direction of lenience to aliens, potentially treating (say) a moderate risk of death as equivalent tó a much greater risk of being beaten up, and treating either as enough to allow the agency to permit an alien to stay in this nation. We need not decide whether there is a real, as opposed to linguistic, difference-between the phrase in the regulation and the phrase in Rodriguez-Molinero. Since Rodriguez-Molinero we have cited both standards interchangeably. See Gutierrez v. Lynch, 834 F.3d 800 , 804-06 (7th Cir. 2016); Velasquez-Banegas v. Lynch, 846 F.3d 258 , 261-62 (7th Cir. 2017); Orellana-Arias v. Sessions, 865 F.3d 476 , 488 (7th Cir. 2017). No decision in this circuit holds the regulation invalid or creates a standard incompatible with it. If there is a difference, it is not one adverse to aliens. By reciting this circuit’s non-quantitative proxy for the regulatory language, the IJ and BIA did not commit a legal error.

The petition for review is denied.