specially concurring.
The majority refuses to approve the consent decree for several reasons. In my view, one of its main reasons is incorrect as a matter of California law, and another is incorrect as a matter of logic and sound public policy; but our decision in Brown v. Ticor Title Insurance Co., 982 F.2d 386 (9th Cir.1992), compels us to reject the consent decree. Accordingly, I concur specially in the result.
A. Actual Damages
A central theme of the majority’s opinion is that the consent decree is unfair because it releases nearly all potential claims by members of the plaintiff class for actual damages. The majority’s underlying premise is incorrect. The representative plaintiff, the defendant, and the district court all agree that the ambiguous consent decree does not release class members’ claims for actual damages. The majority’s antithetical interpretation of the decree contravenes settled law.
In construing a consent decree, we apply the same principles used to interpret a contract. Thompson v. Enomoto, 915 F.2d 1383, 1388 (9th Cir.1990). The rules of contract interpretation of the situs state, here California, govern. Gates v. Rowland, 39 F.3d 1439, 1444 (9th Cir.1994). Under California law, interpretation of a consent decree or a contract begins with an analysis of the agreement’s text. Thompson, 915 F.2d at 1388.
That text supports Defendant’s assertion that the consent decree releases only claims for statutory and treble damages under the Americans with Disabilities Act (ADA) and similar California statutes:
• First, the release applies only to claims “arising under” or “based upon” the cited statutes. According*957ly, all unrelated common law claims — including those for premises liability, intentional infliction of emotional distress, and negligence — as well as all other statutory claims, remain available to potential plaintiffs.
• Second, even with respect to the released claims, the consent decree provides that they “do not include personal injury claims involving physical injury to a plaintiff.” (Emphasis in original.) That clause preserves a large number of claims, including nearly all those for which a plaintiffs damages could exceed the statutory minimum.1
Thus, the wording of the consent decree supports Defendant’s argument that the decree does not release claims for actual damages. However, California law requires us to look beyond the text. We must interpret the consent decree so as “ ‘to give effect to the mutual intention of the parties as it existed at the time of contracting.’ ” Gates, 39 F.3d at 1444 (quoting Thompson, 915 F.2d at 1388 (citing Cal. Civ.Code § 1636)). Although we are to determine the parties’ intention from the words of the consent decree when possible, id., in many circumstances we may also look to other sources to determine what the parties had in mind. For example, when the text of a consent decree is ambiguous, we look to extrinsic evidence to discern the parties’ true intent. S.F. NAACP v. S.F. Unified Sch. Dist., 896 F.2d 412, 414 (9th Cir.1990); United States v. ITT Cont’l Baking Co., 420 U.S. 223, 238 & n. 11, 95 S.Ct. 926, 43 L.Ed.2d 148 (1975).
The text of the consent decree is ambiguous in two ways. First, it is unclear what types of claims are “based upon” the ADA or California disability law. For example, a plaintiff asserting a claim for negligent infliction of emotional distress must prove that the defendant’s breach of a duty imposed by law proximately caused the plaintiff to experience emotional distress. Potter v. Firestone Tire & Rubber Co., 6 Cal.4th 965, 25 Cal.Rptr.2d 550, 863 P.2d 795, 807-08 (1993). Here, the relevant duty would be imposed by statutes requiring that Defendant provide adequate access for disabled persons. Is this type of claim “based upon” the ADA or California disability law and thereby released, or does the release apply only to those claims brought directly pursuant to these statutes? A similar problem is presented by negligence per se claims in which the law violated is the ADA or California disability law. Are such claims preserved by the decree, or are they “based upon” the statutes from which they borrow a duty of care?
A second ambiguity arises from the clause stating that “[t]he Released Claims do not include personal injury claims involving physical injury to a plaintiff.” (Emphasis in original.) Although the majority asserts that this clause unambiguously fails to preserve claims for actual damages arising from emotional distress,2 *958that conclusion ignores two points. First, the preserved claims are all those that “involv[e]” physical injury, not merely those narrower claims “for” physical injury or “based on a” physical injury. Second and relatedly, California courts repeatedly have eschewed the attempt to draw a bright line between physical injury and emotional distress; according to California courts, claims for emotional distress typically “involve” a physical component. See, e.g., Abellon v. Hartford Ins. Co., 167 Cal.App.3d 21, 212 Cal.Rptr. 852, 855 (1985) (“Even doctors have a difficult time distinguishing between ‘mental’ and ‘physical,’ because every emotional disturbance has a physical aspect and every physical disturbance has an emotional aspect.” (citation and internal quotation marks omitted)); Molien v. Kaiser Found. Hosps., 27 Cal.3d 916, 167 Cal.Rptr. 831, 616 P.2d 813, 820-21 (1980) (“[T]he border between physical and emotional injury is not clearly delineated .... In our view ... the attempted distinction between physical and psychological injury merely clouds the issue.”). For these reasons, it is unclear from the text of the release whether the parties intended to foreclose claims for emotional distress or, instead, to preserve them as claims “involving physical injury.”
Because the consent decree is ambiguous in these two ways, California law demands that we go beyond the text of the consent decree to discern the true intent of the parties. Once we do so, the majority’s premise is untenable. The parties’ intent to preserve class members’ claims for actual damages is unmistakably clear. As the majority points out, Defendant repeatedly stated both on the record and publicly that the consent decree does not release claims for actual damages. Maj. op. at 946. The representative plaintiff, too, argues that this was the parties’ intent. Further, the district court, which had a better opportunity to delve into the parties’ intent than an appellate panel does on a cold record, explained that “the consent decree preserves the right of class members to bring claims against ARCO for actual damages.” In other words, all available extrinsic evidence suggests that the parties intended that the decree would preserve class members’ claims for actual damages.
Finally, even if the consent decree were unambiguous, as the majority asserts, California law still would require us to look beyond the face of the consent decree to the true intentions of the parties.3 Under state law, extrinsic evidence is admissible to aid contract interpretation even when the text is clear. Trident Ctr. v. Conn. Gen. Life Ins. Co., 847 F.2d 564, 568 (9th Cir.1988) (noting that “California does not follow the traditional rule.... [It has] turned its back on the notion that a contract can ever have a plain meaning discernible by a court without resort to extrinsic evidence.”). As the California Supreme Court has explained, “[t]he test of admissibility of extrinsic evidence to explain the meaning of a written instrument is not whether it appears to the court to be plain *959and unambiguous on its face, but whether the offered evidence is relevant to prove a meaning to which the language of the instrument is reasonably susceptible.” Pac. Gas & Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 69 Cal.2d 33, 69 Cal.Rptr. 561, 442 P.2d 641, 644 (Cal.1968). So, whether or not the text of the consent decree is ambiguous, California law requires us to consider extrinsic evidence of the parties’ intent. As discussed above, that evidence clearly establishes that the decree was not meant to release claims for actual damages.
In summary, I would give effect to the parties’ intentions and hold that the consent decree does not release class members’ claims for actual damages.
B. Adequate Representation
I do not share the majority’s suspicion that the consent decree in this case was a product of collusion. The majority’s inference rests on two untenable grounds.
First, the majority relies on its view that “the ultimate terms of the decree ... waived practically all of the class members!”] claims without compensation.” Maj. op. at 956. As explained above, class members’ claims for actual damages were not waived.
Second, the majority faults the parties for reaching “an agreement regarding the primary components of the consent decree within four months.” Maj. op. at 955. Early dispute resolution is salutary, and we should not encourage the unnecessary expense, delay, and uncertainty caused by lengthy litigation when the parties are prepared to compromise. Nor should we hold, as the majority does, maj. op. at 956, that a prompt settlement necessarily suggests a failure to prosecute or defend the action with due diligence and reasonable prudence. To the contrary, an early resolution may demonstrate that the parties and their counsel are well prepared and well aware of the strengths and weaknesses of their positions and of the interests to be served by an amicable end to the case.
In summary, I would hold that the district court did not abuse its discretion in finding that the named plaintiff and his counsel fairly and adequately protected the interests of the class, as required by Federal Rule of Civil Procedure 23(a)(4).
C. Treble Damages
Under our precedent, statutory treble damages may be significant enough to require that each class member have an opportunity to opt out of the litigation, as a matter of due process. Brown, 982 F.2d at 387, 392. Here, class members did not have that opportunity, but the settlement released statutory minimum and treble damages, which are significant. For that reason, we have no choice but to reverse and remand.
For this reason, I concur in the judgment.
. The majority cites claims for "property damages” as one type of claim that would be released under this clause. Maj. op. at 1578. I disagree for two reasons. First, many claims for property damage could be brought as common law claims, for example based on theories of negligence or premises liability. Second, most cases in which significant property damage occurs are'likely to "involv[e]” physical injury.
. Under California law, a claim for emotional distress is classified as a personal injury claim. See, e.g., Cantu v. Resolution Trust Corp., 4 Cal.App.4th 857, 6 Cal.Rptr.2d 151, 170 (1992) (holding that "intentional infliction of emotional distress is an injury to the *958person”; Billmeyer v. Plaza Bank of Commerce, 42 Cal.App.4th 1086, 50 Cal.Rptr.2d 119, 125 (1995) (holding that claims for intentional and negligent infliction of emotional distress are barred by the one-year statute of limitations for "personal injury actions”).
. But see Gates v. Rowland, 39 F.3d 1439, 1444 (9th Cir.1994) (suggesting that the Ninth Circuit, applying California law, does not examine extrinsic evidence if the text of the consent decree is clear); Thompson v. Enomoto, 915 F.2d 1383, 1388 (9th Cir.1990) (same).' These cases are unpersuasive on this particular point, however, because they are internally inconsistent. Although they purport to apply California contract law, they misconstrue state law regarding the admissibility of extrinsic evidence.