dissenting.
In the principal suit, I respectfully dissent. I would hold that the district court erred (1) in not applying collateral estoppel against State Farm and (2) in granting summary judgment in favor of State Farm. In the independent action, I would reinstate the action because the majority construes the circumstances in which a party commits fraud on the court too narrowly. If proved, the agents’ allegations against Orrick and State Farm amount to fraud on the court sufficient to support an independent action.
I. Breach of Contract Action
A. Collateral Estoppel
The majority holds that collateral estop-pel does not bar relitigation in this case of the identical issue that was decided in Sandberg v. State Farm Mut. Auto. Ins. Co., 182 F.3d 927 (9th Cir.1999) (table) (memorandum disposition). Sandberg held that, under California law, the contract provisions that are at issue here permit termination only for good cause. The majority concedes that the district court erred in holding that the preconditions for estoppel were not met in this case, see majority opinion, supra, but the majority nonetheless defers to the district court’s ultimate decision not to estop State Farm from relitigating the issue. The majority errs.
To decide as it does, the majority must rely on the exception that “[ajllowing offensive collateral estoppel may ... be unfair to a defendant if the judgment relied upon as a basis for the estoppel is itself inconsistent with one or more previous judgments in favor of the defendant.” Parklane Hosiery Co., v. Shore, 439 U.S. 322, 331, 99 S.Ct. 645, 58 L.Ed.2d 552 (1979) (emphasis added). We do not apply a mechanical approach to the question of when prior inconsistent judgments preclude the application of collateral estoppel. Instead, we have held that Parklane Hosiery is a guide to determining when the application of collateral estoppel may be unfair to a defendant. See Robi v. Five Platters, Inc., 838 F.2d 318, 330 (9th Cir.1988) (“We prefer to read Parklane as providing guidance to the district courts for the appropriate exercise of discretion as to when to apply offensive issue preclusion in the presence of inconsistent judgments.”).
Although Sandberg decided that State Farm had good cause to terminate the agent in that case, it held that under California law the same termination clause that is at issue here permitted termination only for good cause. The majority’s conclusion that Sandberg is “inconsistent” with other authority so as to preclude applying collateral estoppel against State Farm is simply wrong. To be inconsistent in this sense, a decision must conflict with relevant legal authority on the same issue.
This case involves two groups of plaintiffs: (1) those from Arizona, California, Colorado, Iowa, Maryland, Nevada, Utah, and Washington — states that the district court held follow a rule that permits extrinsic evidence to be used to illuminate the meaning of unambiguous contract terms; and (2) those plaintiffs from Florida, Idaho, Kansas, Louisiana, Nebraska, Ohio, and Texas — states that the district court held permit the introduction of extrinsic evidence if a contract term is ambiguous. Neither party has challenged *782the district court’s conclusions in this regard. Accordingly, the only authority that would be relevant in determining that Sandberg was “inconsistent” would need to be authority interpreting the rule of law that we applied in Sandberg — that is, authority applying California law, or the law of a state that follows the same rule.
The pre-Sandberg authorities that State Farm proffers applied New Hampshire law, Illinois law, New Mexico law, and Kansas law — none of which is at issue here (except Kansas). Although State Farm cites three cases from California courts, two are from trial courts, and none of them is California Supreme Court authority. It is wrong to hold, on such a basis, that Sandberg is “inconsistent” with the law of California and of the other states that follow the same rule. Nor do any of the authorities that post-date our decision in Sandberg indicate that Sandberg is inconsistent.
Moreover, the majority is wrong to defer to the district court’s decision in disregarding Sandberg. The district court has no such discretion, and we owe the district court’s decision on this point no deference. The district court held that our decision in Sandberg was sufficiently inconsistent with prior decisions in other jurisdictions (not our court) that it would be unfair to estop State Farm from relitigating the scope of the termination clauses at issue. In such circumstances, the lower court’s refusal to follow a decision of this circuit is reversible error.
The question of the consistency of Sand-berg with prior decisions — and, hence, in some sense Sandberg’s fairness — was a matter for the panel that decided that case. It is improper to assume that the Sandberg panel was unaware of, or disregarded, the state of the law when it issued its disposition. Moreover, State Farm had ample opportunity to challenge the consistency (and fairness) of our decision with prior authority when it petitioned for rehearing by the original panel. State Farm again had the chance to raise the issue in its petition for certiorari that was denied by the Supreme Court. See Sandberg v. State Farm Mutual Auto. Ins. Co., 528 U.S. 1118, 120 S.Ct. 938, 145 L.Ed.2d 816 (2000) (denying petition for certiorari). I submit that the district court here had no discretion to refuse to follow our Sandberg decision because, in the lower court’s judgment, our decision did not agree with prior judgments from different courts.1
Nothing in Parklane Hosiery suggests that the Supreme Court intended to upset the usual channels of appellate review or to grant discretion to district courts to refuse to accord preclusive effect to prior judgments of their circuit courts. Nor does Parklane Hosiery suggest that we should defer to an inferior court within our jurisdiction as to the correct legal application of our own decisions. Parklane Hosiery involved the question of whether to grant preclusive effect to a decision by the Second Circuit in a ease that was before a district court within the same circuit. See id. at 324-25, 99 S.Ct. 645. But the Supreme Court did not reach the actual question presented here of whether a district court may disregard a prior decision by the court of appeals that sits above it on the ground that the circuit’s decision conflicts with foreign authority. Although the Court noted that a general exception might be made for decisions that are inconsistent with prior decisions, it held that *783“none of the circumstances that might justify reluctance to allow the offensive use of collateral estoppel is present.” Id. at 331, 99 S.Ct. 645.
The district court has no special expertise nor license to assess, as a matter of law, whether our decisions are right, wrong, consistent, or inconsistent. That is, quite simply, our task (and that of the Supreme Court). Our decisions, whether memorandum dispositions or opinions, state the law of the circuit and must be followed by a district court in cases where they properly govern the issues presented. See Ninth Circuit Rule 36-3. The district court has no authority to disregard our decision.
In short, the district court’s decision that State Farm is not collaterally es-topped from relitigating whether the contracts here are terminable for cause was reversible error. The district court may not disregard our decision, and the conclusion that Sandberg is inconsistent with the weight of authority from other jurisdictions on the questions presented in this case is inaccurate.
B. Summary Judgment
I also disagree with the majority that summary judgment was proper. I would reverse and remand for trial. In affirming the grant of summary judgment, the majority ignores the settled standards for summary judgment and substitutes its own interpretation of the evidence for one that views the evidence in the light most favorable to the nonmoving party. At summary judgment, the court is prohibited from taking such a view of the evidence. See Mat-sushita Elec. Indus. Co. v. Zenith Radio Corp., 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986).
Neither party disputes that extrinsic evidence may be used to illuminate and interpret the meaning of the contract terms that are at issue here — at least as to the plaintiffs who have stated claims for breach of contract under the law of states, such as California, that permit such evidence to be used to interpret unambiguous contract terms. See, e.g., Pacific Gas and Elec. Co. v. G.W. Thomas Dray age & Rigging Co., 69 Cal.2d 33, 69 Cal.Rptr. 561, 442 P.2d 641, 644-47 (1968) (en banc). The plaintiffs allege that the deletion of the “at will” provision from a previous version of the agency agreement, the addition of a termination review procedure, and State Farm documents that state that State Farm would not terminate agents except for serious infractions require a holding that the contract permits termination only for cause. Read together, this evidence, considered in the light most favorable to the plaintiffs, clearly supports the reasonable inference that the agency contracts were terminable only for cause.2 At summary judgment the plaintiffs need show no more in order to proceed to trial.
The majority is wrong to hold that what the plaintiffs ask us to do is to write in or “graft” onto the contract a “for cause” provision. All that is required for the plaintiffs to prevail at this stage of the case is to present evidence (that we must view in the fight most favorable to the plaintiffs) that raises a disputed issue of material fact as to the meaning of the contract. That they have done.
*784Of course, because the contracts, considered in the light most favorable to the plaintiffs, reasonably may be read to allow termination only for cause, the district court’s rulings that State Farm breached no implied covenants of good faith and fair dealing as to the agents’ other claims should be reversed as well.
II. The Independent Action
The gravamen of the plaintiffs complaint in the independent action under Rule 60(b) is that State Farm’s counsel, Orrick, Herrington & Sutcliffe, committed fraud upon the court by responding to the plaintiffs’ request to depose Henry Keller, a former State Farm executive, by fraudulently claiming to represent Keller and opposing the deposition on grounds that he was very elderly, had long been retired, and possessed no documents or knowledge concerning the subject matter of the lawsuit. In reliance on Orrick’s representations, the plaintiffs did not pursue the deposition. In reality, the plaintiffs claim, Orrick did not represent Keller and knew Keller possessed material evidence that strongly supported their claims and was willing and ready to testify.
The majority holds that Orrick’s acts are not actionable because they do not rise to the level of fraud on the court or work a grave miscarriage of justice. I cannot agree. The majority’s interpretation as to when a federal court may set aside a judgment is too narrow. Rule 60(b) permits a court, within a reasonable time, not more than one year after a judgment is entered, to grant a motion for relief from a judgment because of, inter alia, fraud on the court. See Fed.R.Civ.P. 60(b). After a year elapses, the rule also allows a party to bring an independent action to set aside a judgment for fraud on the court. The power to grant relief in an independent action is based in the court’s inherent “power to vacate judgments on proof that a fraud on the court has been committed.” In re Levander, 180 F.3d 1114, 1118 (9th Cir.1999) (citing Chambers v. NASCO, Inc., 501 U.S. 32, 43-44, 111 S.Ct. 2123, 115 L.Ed.2d 27 (1991)). Rule 60(b) largely codified preexisting substantive law regarding the circumstances in which courts are empowered to set aside judgments, but removed many procedural complexities that were associated with preexisting law. See Averbach v. Rival Mfg. Co., 809 F.2d 1016, 1020-21 (3d Cir.1987) (discussing the relationship of Rule 60(b) to prior law and practice). The period of a year specified by Rule 60(b) supplanted the earlier requirement that a request to set aside a judgment that was brought as part of the original action must have been made before end of the term of court in which the judgment was entered. However, nothing in the text of Rule 60(b) requires that a plaintiff make a higher showing for relief in an independent action than in a motion under the rule.
The question presented here is whether the fraud that the plaintiffs have alleged may, if proved, warrant relief. Fraud in the discovery process plainly may sustain an independent action under Rule 60(b). The Supreme Court has held that federal courts may set aside judgments, notwithstanding the defense of laches or the plaintiffs lack of diligence in pursuing relief if a judgment has been obtained based on evidence that counsel has “manufactured.” See Hazel-Atlas Glass Co. v. Hartford-Empire Co., 322 U.S. 238, 64 S.Ct. 997, 88 L.Ed. 1250 (1944). The Third Circuit has held that a plaintiff stated a claim for relief in an independent action based on opposing counsel’s false and misleading answers to interrogatories. See Averbach, 809 F.2d at 1017, 1022-23. None of our precedents warrants a different result. In Levander, we stated that “non-disclosure by itself does not constitute fraud on the court” that warrants relief in an independent action under Rule 60(b), Levander, 180 F.3d at 1119, and that ordinary perjury similar*785ly is not actionable, but we explicitly embraced a definition of fraud on the court that encompasses “fraud perpetrated by officers of the court so that the judicial machinery cannot perform in the usual manner its impartial task of adjudging cases.” Id. at 1119 (quoting 7 James W. Moore, et al., Moore’s Federal Practice ¶ 60.33, at 515 (2d ed.1978)).
Here, Orrick is alleged not merely to have failed to disclose material evidence3 that under long-established federal law it was obligated to disclose, but also to have claimed to represent a potential witness that it did not represent and intentionally to have misrepresented that person’s position in a calculated attempt to obtain a judgment in its client’s favor. An attorney in a legal proceeding who fraudulently claims to represent a client whom she or he does not, not only commits a fraud against the parties to the proceeding, but also a fraud on the court. Both the court and the parties are entitled to rely on an attorney’s representation that she or he actually represents a client whom he or she claims to represent. A fraudulent misrepresentation about the fact, nature, and scope of an attorney’s mandate is not only an ethical breach but a fraud of the highest order that interferes with the machinery of our judicial system. Our adversary system cannot function if attorneys claim to represent individuals that they do not. The majority fails to appreciate the gravity of Orrick’s alleged fraud, and errs as a matter of law in holding that an allegation of such fraud may be dismissed for failure to state a claim.
The majority also errs by holding that the plaintiffs could have, with diligence, discovered Orrick’s alleged duplicity by insisting on deposing Keller despite Orrick’s representations and that the plaintiffs’ failure to do so bars their claim. I strongly disagree. First, under Hazel-Atlas, the diligence of a party in discovering the alleged basis for setting aside a judgment is not decisive. See Hazel-Atlas, 322 U.S. at 246, 64 S.Ct. 997. Moreover, unlike the perjury in Levander, Orrick’s alleged fraud will not be subject to the rigors of cross-examination in the factfinding process itself: The representation by counsel that he represents a witness and that the witness has no material information is not “testimony” that may be cross-examined. A party is entitled to rely on the representations of adverse counsel regarding such matters.
As a matter of construing the discovery rules, moreover, the principle that the majority embraces, namely that a party may not rely on the plain representations of adverse counsel in the discovery process, is not only cynical but unworkable. It requires a party doggedly to pursue discovery notwithstanding opposing counsel’s signed certification that such discovery is not warranted, see Fed.R.Civ.P. 26(g), on the possible chance that opposing counsel is committing fraud. The failure of the party seeking discovery to take such action, which itself might violate the discovery rules, by the majority’s lights, precludes it from seeking redress if it later discovers that it has been the victim of fraud. I cannot embrace such a view of the law.
Such an interpretation of the discovery rules is not only unconscionable, but flies in the face of the purpose of the rules themselves. As the Rules Advisory Committee has explained, Rule 26 explicitly imposes an affirmative duty “to engage in pretrial discovery in a responsible manner that is consistent with the spirit and purposes of Rules 26 through 37,” Fed. R.Civ.P. 26, Advisory Committee Notes, 1983 Amendment, Subdivision (g), and Rule 37(c) permits a district court to sanction a party for making false or misleading *786discovery disclosures. See Fed.R.Civ.P. 37(c). As the Washington Supreme Court has held, the Federal Rules of Civil Procedure explicitly encourage the imposition of sanctions for discovery abuse in part because “a spirit of cooperation and forthrightness during the discovery process is necessary for the proper functioning of modern trials.” Wash. State Physicians Ins. Exch. & Ass’n v. Fisons Corp., 122 Wash.2d 299, 858 P.2d 1054, 1077 (1993). The court noted that although “[f]air and reasoned resistance to discovery is not sanctionable[,] ... misleading ... responses [are] ... contrary to the purposes of discovery and ... most damaging to the fairness of the litigation process.” Id. at 1079-80.
The plaintiffs’ allegations, if proved, would be most damaging to the ethical standing and professional reputation of any attorney — and indeed might have severe collateral consequences such as disbarment.4 The majority minimizes the gravity of the plaintiffs’ charge and invites similar behavior in the future by holding that such a claim may be dismissed for failure to state a claim in an independent action under Rule 60(b). As Judge Schwarzer has persuasively explained: “Misconduct, once tolerated will breed more misconduct and those who might seek relief against abuse will instead resort to it in self-defense.” William W Schwarzer, Sanctions Under the New Federal Rule 11 — A Closer Look, 104 F.R.D. 181, 205. I would reverse the district court and reinstate the action.
. The district court considered only authority prior to Sandberg in concluding: “The Court ... takes note of the stark landscape of inconsistent prior interpretations of the termination clause, and finds that this factor weighs heavily in favor of granting Sandberg no preclusive effect.” The district court’s decision amounts to nothing more than an expression of disagreement with the correctness of the Sand-berg decision.
. The district court dismissed much of the evidence that the plaintiffs proffered because it was written after the contract provisions at issue were drafted. The court reasoned that because the evidence postdated the drafting process it lacked any nexus to the drafting and could not illuminate the meaning of the contract provisions. Although such an inference, if justified, could be drawn after trial, it is not permissible at the summary judgment stage. It is entirely possible that the materials that were written after the contract provisions are evidence of what the parties to the contract intended by its provisions.
. Keller himself advised Orrick that he had material evidence to furnish to the plaintiffs.
. The majority does not hold that Orrick and State Farm have not committed the actions as alleged, but merely that the plaintiffs may not seek relief under Rule 60(b). Orrick itself could be well-served by trial of the charges. They must be very disturbing to a reputable firm. If untrue, Orrick should welcome the opportunity to disprove them.