(Concurring in part, Dissenting in part)..
Although I concur with the majority regarding the injunctive class, I respectfully dissent from the discussion of the state damages classes.30 In my view, the opinion erodes the discretion which we are required to afford to the district court in class certification proceedings. A district court’s decision to certify a class is reviewed under the deferential abuse of discretion standard. See Smilow, 323 F.3d at 37 (“Orders certifying or decertifying a class are reviewed for abuse of discretion.”) (citing Califano v. Yamasaki, 442 U.S. 682, 703, 99 S.Ct. 2545, 61 L.Ed.2d 176 (1979)); see also Blyden v. Mancusi, 186 F.3d 252, 269 (2d Cir.1999) (“A district court’s decision to certify a class is reviewed for abuse of discretion, and ‘[a] reviewing court must exercise even greater deference when the district court has certified a class than when -it has declined to do so.’ ’’(citation omitted)).
Rule 23 grants the district court broad discretion to determine whether a class should be certified. Fed.R.Civ.P. 23. Our review is, therefore, focused on whether the district court properly applied the criteria set out in Rule 23. See Mowbray, 208 F.3d at 295 (“An abuse occurs when a court, in making a discretionary ruling, relies upon an improper factor, omits consideration of a factor entitled to substantial weight, or mulls the correct mix of factors but makes a clear error of judgment in assaying them.”). Importantly, a district court remains free at a later stage to modify or even decertify a class if later evidence disproves the plaintiffs’ assertions regarding, for example, the predominance of common issues. See Falcon, 457 U.S. at 160, 102 S.Ct. 2364 (“Even after a certification order is entered, the judge remains free to modify it in the light of subsequent developments in the litigation.”); In re Visa Check/MasterMoney Antitrust Litig., 280 F.3d 124, 141 (2d Cir.2001).
In this case, the district court addressed all of the Rule 23 requirements: numerosity of the class members; commonality of the questions of law or fact; typicality of the claims or defenses; adequacy of representation; predominance of common questions; and the superiority of the class action as an adjudicative vehicle. And, as required in this circuit, the district court “conduct[ed] a rigorous analysis of the prerequisites established by Rule 23,” Smi-low, 323 F.3d at 38, and “formulate[d] some prediction as to how specific issues wfould] play out in order to determine whether common or individual issues predominate,” Mowbray, 208 F.3d at 298.
The majority does not question the rigor with which the court conducted its analysis. Indeed, the opinion applauds the court’s “attemptf ] to meet its obligations ... to conduct a rigorous analysis at the certification stage.” Slip op. at 26. Rather, the basis for the majority’s remand on the certification of the damages classes is the insufficiency of evidence available to the district court. Although the opinion faults the timing of the certification (and, thus the incompleteness of the record), I am concerned that in vacating and remanding the certification order for reconsideration with additional evidence from the plaintiffs, the opinion stands for the *31proposition that we now require a high level of fact-finding before certification.
I agree with the district court that our case law does not “mandat[e] a particular level of factfinding by the district judge at the certification stage.” Motor Vehicles VI, 241 F.R.D. at 82 n. 7. On the contrary, in PolyMedica, we stated that:
The question of how much evidence ... is necessary for a court to accept the [theory of reliance] at the class certification stage is therefore one of degree. District courts must draw these lines sensibly .... We have no illusions that this line-drawing is easy. Knowing the high stakes in the class-certification decision, the parties will try to move the court in different directions, with plaintiffs arguing for less evidence ... and defendants for more ... the district court must evaluate the plaintiffs evidence ... critically without allowing the defendant to turn the class certification proceeding into an unwieldy trial on the merits.
432 F.3d at 17 (emphasis added). The district court drew those lines sensibly in this case. We are not entitled to second-guess that decision in the absence of evidence that it engaged in an abuse of discretion. Our decision to vacate and remand the certification of the damages classes to allow the district court the benefit of full discovery, effectively overrides the district court’s assessment of how much evidence it needed to certify the class. Under the banner of a “novel and complex” theory in a class certification proceeding, we now appear to require district courts to fully vet and test the underpinnings of a plaintiffs legal theory. See slip op. at-(“[I]n our view, a searching inquiry is in order where there are not only disputed basic facts, but also a novel theory of legally cognizable injury.”).
At issue in this case are questions regarding the plaintiffs’ theory of impact. Although the district court admitted some concern about whether the plaintiffs’ proof of impact would be “sufficient to withstand a motion for summary judgment or for judgment as a matter of law at trial,” Motor Vehicles V, 235 F.R.D. at 139, the district court properly remained focused on the certification requirements and concluded that “the plaintiffs’ proof does meet the commonality and typicality standard.” Id. The majority even admits that these questions regarding the plaintiffs’ theory of impact “are themselves susceptible to common proof (the potential size of the gray market and the distinction between the effects of horizontal and vertical restraints).” Slip op. at 28 n. 29. The opinion goes on, however, to conclude that a more searching inquiry into the factual basis of that theory is required here. I disagree.
In this case, the questions regarding the viability of the plaintiffs’ theory of impact are not limited to certification issues, but go to the merits of the plaintiffs’ claim. I disagree with the majority’s reading of PolyMedica and Xcelera, which they cite in support of their position that “factual bases of theories of common proof are appropriately, although preliminarily, tested at the class certification stage.” Id. PolyMedica and Xcelera were securities class actions in which the plaintiffs sought to use the fraud-on-the-market presumption to demonstrate market efficiency. Those class actions were brought under § 10(b) of the Securities Exchange Act of 1934, and Rule 10b-5 promulgated thereunder, which require that ea$i plaintiff prove that he or she individually relied on the alleged misrepresentation. A requirement of individualized reliance “would effectively preclude securities fraud class actions under Rule 23(b)(3) [because] [individual issues of reliance *32would necessarily overwhelm the common ones.” Xcelera, 430 F.3d at 507. The applicability of the fraud-on-the-market theory is central to the appropriateness of the class action as a vehicle for litigation: under the theory, plaintiffs are no longer required to prove individualized reliance. See PolyMedica, 432 F.3d at 18-19 (vacating and remanding the class certification after concluding that the district court had committed error in adopting the incorrect definition of “market efficiency,” one of the elements for invoking the fraud-on-the-market presumption). In this antitrust case, the majority points to no legal error committed by the district court in assessing the appropriateness of certification.
In remanding the certification of the damages classes for reconsideration with the benefit of additional evidence, the majority conflates the dispute as to the viability of the plaintiffs’ theory with the specific inquiries required at class certification. As we noted in PolyMedica, “a court has the power to test disputed premises early on if and when the class action would be proper on one premise but not another.” PolyMedica, 432 F.3d at 6 (quoting Tar-diff, 365 F.3d at 4-5) (emphasis added). Indeed, insofar as those premises are not preclusive of the class action as a vehicle, we have no basis for requiring a district court to inquire further into the merits of the case. While the plaintiffs’ theory of antitrust impact is novel and complex, it, unlike the fraud-on-the-market theory, is not determinative of whether a class action is proper or not. Indeed, the identified uncertainties within the plaintiffs’ theory challenge only the ability of the plaintiffs (as a group) to successfully prove their theory of impact. Slip op. at 26-28. In this case, an inquiry that tests each stage of the plaintiffs’ theory is, in effect, an assessment of the case’s merits. As such, we are putting the cart before the horse and turning the class certification stage into a motion for summary judgment proceeding — the appropriate juncture at which to fully vet the viability of the plaintiffs’ theory. In so holding today, I fear that we are removing the underpinnings of the discretion we grant district courts to draw sensible lines and further blurring the distinction between the certification inquiry and a trial on the merits. I believe this course is erroneous and contrary to established precedent, and thus dissent.
. I agree that the case is properly remanded to the district court so that it can first establish whether there is jurisdiction under § 1332 or § 1367.