OVERVIEW

Petitioner Steffany Tremain (“Tremain”) brought suit against Bell Industries, Inc. (“Bell Industries”); Bell Industries, Inc. Long Term Disability Plan (“the Bell Plan”); and Metropolitan Life Insurance Company (“MetLife”) (collectively, “the Defendants”). Tremain claimed the Defendants wrongfully terminated her long-term disability payments under the Employee Retirement Income Security Act of 1974 (“ERISA”), 29 U.S.C. § 1001 , et seq. In addition, Tremain asserted claims for breach of fiduciary duty under 29 U.S.C. §§ 1105 , 1109(a) and 1132(a)(2) and claims for current and retroactive benefits under 29 U.S.C. § 1132 (a)(1)(B). The district court granted the Defendants’ motion for summary judgment as to all of Tremairis claims. Tremain appeals the district court’s judgment only as to her claims under 29 U.S.C. § 1132 (a)(1)(B). 2

Tremain contends the district court erred in applying the arbitrary and capricious standard to review MetLife’s decision to terminate her long-term disability benefits. Tremain also contends the district court erred in refusing to consider evidence outside the administrative record. We have jurisdiction under 28 U.S.C. § 1291 . We reverse the district court’s summary judgment, and remand for a trial in which the district court is directed to review de novo the termination of Tremain's benefits under the Bell Plan, and determine the amount of any unpaid benefits that might be due to her.

BACKGROUND

Tremain was a General Sales Manager for Bell Industries. As of March 1, 1990, chronic and severe pain caused her to become unable to work. She filed a claim for disability benefits with MetLife, the insurer and claims administrator of the Bell Plan. On August 28, 1990, she began receiving benefit payments in the amount of $4,250 per month. Over the next four- and-one-half years, Tremain remained under the care of several treating physicians, each of whom confirmed her ongoing disability.

In January 1995, MetLife requested a medical case management review of Tre-, main’s medical records by Susan Yager, R.N., a Rehabilitation Nurse and Medical Management Specialist. Based on Ms. Yager’s recommendations, Tremain entered the Stanford University In-Patient Pain Management and Chemical Dependency Program. After Tremairis discharge from the Stanford Pain Clinic, one of its physicians, Dr. John Massey, continued to treat her on an outpatient basis.

In September 1995, Dr. Massey submitted a “Physical Capacities Evaluation” to MetLife stating that in an eight-hour workday Tremain was able to sit for five hours, stand for two hours and walk for one hour. On September 8, 1995, Dr. Periroth, another of Tremairis treating physicians, submitted a Physical Capacities Evaluation to MetLife with conclusions almost identical to Dr. Massey’s.

Based on these findings, MetLife ordered a vocational assessment analysis of Tremain. After considering the analysis, MetLife determined that Tremain could work in a capacity similar to her previous employment. In November 1995, MetLife employed Dr. Robert Petrie, a specialist in occupational medicine, to review Tremairis medical records. After reviewing Tre-mairis file, Dr. Petrie concluded that Tre-main was not totally disabled.

On December 6, 1995, MetLife requested an additional vocational review. The *974 review concluded that suitable jobs involving work Tremain could perform had median annual salaries of $36,908 to $49,072. In addition, the analysis found that such annual wages varied between $25,000 and $250,000 “depending on the manager’s level of education, experience, industry, and the number of employees he or she supervises.”

On January 31, 1996, MetLife notified Tremain that it would terminate her disability benefits effective February 29, 1996, based on its finding that she no longer met the plan’s definition of “Total Disability.” Further, the letter reiterated the findings in the vocational assessment, stating that annual wage estimates for occupations in her field ranged from $36,908 to $49,072.

Tremain appealed MetLife’s termination of her disability benefits, and asked for a review and reversal of its decision. Tre-main informed MetLife that the definition of “Total Disability” listed in MetLife’s termination letter was the definition in the Bell & Howell Long Term Disability Plan (“Bell & Howell Plan”), not the Bell Plan in which she was a participant. The Bell Plan’s definition of “Total Disability” had more generous provisions. 3 While both plans provided for long term total disability if, after 24 months of benefit payments, the participant could not perform the duties of any work for which she was qualified, the Bell Plan also provided that if a participant’s earning capacity decreased by fifty percent, then she was considered “Totally Disabled.”

Besides informing MetLife that it had applied the wrong definition, Tremain informed MetLife that her 1989 W-2 Form listed her annual earnings as $108,731. She contended that because the highest wage indicated in MetLife’s termination letter was $49,072, her earning capacity had decreased by at least fifty percent, which meant she was totally disabled according to the Bell Plan’s alternative method for determining total disability.

In addition to appealing the termination of her benefits, Tremain sought a review of the benefits she had already received from MetLife. Tremain contended that Met-Life used the Bell & Howell Plan to calculate her monthly benefits instead of the Bell Plan, which granted greater monthly benefits for salespersons. 4 As a result, Tremain claimed MetLife underpaid her $2,640 per month for each of the five-and-one-half years she received disability benefits.

*975 Tremain supplemented her appeal with additional documentation from her treating physicians. Two of her physicians, Dr. Perlroth and Dr. Margóles, certified to MetLife that Tremain was unable to work. MetLife forwarded to Dr. Petrie Tremain’s supplemental medical record, which included Dr. Perlroth’s and Dr. Margoles’s most recent diagnoses, and in which Dr. Mar-góles stated that “[a]t the present time, I rate [Tremain’s] disability as permanent.” Dr. Petrie nevertheless concluded that Tremain’s record demonstrated that her medical condition had improved and she was able to return to work. MetLife refused to reinstate Tremain’s disability benefits or recalculate the monthly benefits already paid.

Tremain then filed suit in the district court against Bell Industries, the Bell Plan, and MetLife, asserting claims for breach of fiduciary duty under 29 U.S.C. §§ 1105 , 1109(a) and 1132(a)(2) and claims for current and retroactive benefits under 29 U.S.C. § 1132 (a)(1)(B). The district court granted summary judgment in favor of the Defendants as to Tremain’s breach of fiduciary duty claims, holding that ERISA precluded relief under sections 1109(a) and 1132, which only allow recovery for the benefit of the plan as a whole, not for individual participants. Regarding Tremain’s claims for current and retroactive benefits under 29 U.S.C. § 1132 (a)(1)(B), the district court granted summary judgment in favor of MetLife and Bell Industries, determining that under § 1132(a)(1)(B) a plaintiff may only recover benefits against a plan.

As to the Bell Plan, the district court held that it would review MetLife’s decision to terminate Tremain’s benefits by an arbitrary and capricious standard 5 because the Bell Plan’s provision that benefits would be paid upon evidence “satisfactory to us” gave MetLife discretion to determine eligibility and to interpret the terms of the plan. The district court declined to apply a de novo standard of review to MetLife’s benefits decision even though MetLife had a conflict of interest, because it concluded Tremain had failed to present material evidence that MetLife’s conflict affected its decision to deny her benefits. In making its decision as to what standard of review to apply, the district court refused to consider any evidence that was not part of the administrative record, such as W-2 statements, Tremain’s employment agreement, and the deposition testimony of a Bell employee, Louis Weis. The district court also refused to consider such evidence when it reviewed MetLife’s decisions to terminate Tremain’s benefits and to deny her claim that she had been underpaid. Applying an abuse of discretion standard, the district court granted summary judgment, upholding MetLife’s benefits decisions and dismissing Tremain’s claims under 29 U.S.C. § 1132 (a)(1)(B). This appeal followed.

ANALYSIS

We review de novo a district court’s grant of summary judgment. See Lang v. Long-Term Disability Plan of Sponsor Applied Remote Technology, Inc., 125 F.3d 794, 797 (9th Cir.1997). We also “review de novo the district court’s choice and application of the standard of review applicable to decisions by fiduciaries in the ERISA context.” Id. (citing Taft v. Equitable Life Assurance Soc’y, 9 F.3d 1469 , 1471 (9th Cir.1993)). We must determine, viewing the evidence in the light most favorable to the nonmoving party, here *976 Tremain, whether there are any genuine issues of material fact and whether the district court correctly applied the relevant substantive law. See Margolis v. Ryan, 140 F.3d 850, 852 (9th Cir.1998).

We review a district court’s evi-dentiary rulings for an abuse of discretion. See Russian River Watershed Protection Comm. v. City of Santa Rosa, 142 F.3d 1136 , 1144 n. 6 (9th Cir.1998). In particular, a district court may abuse its discretion if it does not apply the correct law. See United States v. Sprague, 135 F.3d 1301, 1304 (9th Cir.1998); see also Koon v. United States, 518 U.S. 81, 100 , 116 S.Ct. 2035 , 135 L.Ed.2d 392 (1996) (“A district court by definition abuses its discretion when it makes an error of law.”).

We review de novo an ERISA plan administrator’s decision to deny benefits “unless the benefit plan gives the administrator or fiduciary discretionary authority to determine eligibility for benefits or to construe the terms of the plan.” Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 115 , 109 S.Ct. 948 , 103 L.Ed.2d 80 (1989). When discretion is conferred, we generally review the exercise of that discretion for abuse of discretion. See Winters v. Costco Wholesale Corp., 49 F.3d 550, 552 (9th Cir.1995) (quoting Taft, 9 F.3d at 1471).

If, however, the plan administrator is also the insurer “that conflict [of interest] must be weighed as a ‘facto[r] in determining whether there is an abuse of discretion.’ ” Snow, 87 F.3d at 330 (quoting Firestone, 489 U.S. at 115 , 109 S.Ct. 948 ). Our review in such a circumstance, although still for abuse of discretion, is “less deferential.” Lang, 125 F.3d at 798 (citing Snow, 87 F.3d at 331). If, however, the program participant presents “material, probative evidence, beyond the mere fact of the apparent conflict, tending to show that the fiduciary’s self interest caused a breach of the administrator’s fiduciary obligations to the beneficiary,” Lang, 125 F.3d at 798 (quoting Atwood v. Newmont Gold Co., 45 F.3d 1317, 1322 (9th Cir.1995)), a rebuttable presumption arises in favor of the participant. The plan then “bears the burden of rebutting the presumption by producing evidence to show that the conflict of interest did not affect its decision to deny or terminate benefits.” Id. If the plan fails to carry this burden of rebutting the presumption, we review de novo its decision to deny benefits. See id. (quoting Atwood, 45 F.3d at 1323 ).

In this case, MetLife both funds the Bell Plan and acts as the plan’s administrator. There is, therefore, a conflict of interest. As a result, we must determine whether there is sufficient material and probative evidence to trigger a rebuttable presumption that MetLife’s conflict of interest affected its decision to deny Tre-main benefits. In making this determination, the first issue we confront is what evidence a district court may consider in deciding whether the rebuttable presumption has been established, and in deciding whether the plan has rebutted that presumption.

The district court, relying on Taft, held that its “review of the evidence is limited to the administrative record.” In Taft, the district court considered a doctor’s testimony that had not been before the plan administrator at the time it denied the plaintiff benefits. See Taft, 9 F.3d at 1471. On appeal we held that the district court erred in considering evidence outside the administrative record. See id. at 1472. We reasoned that allowing the court to consider evidence that was not before the plan administrator, when reviewing the plan administrator’s benefits decision for abuse of discretion, would lead “to the anomalous conclusion that a plan administrator abused its discretion by failing to consider evidence not before it.” Id.

In the present case, Tremain offered her evidence from outside the administrative record on a different issue. The issue she presented was whether the plan *977 administrator’s conflict of interest affected its decision to deny her benefits. That is a threshold issue which must be decided before a court can determine what standard of review to apply to a plan administrator’s benefits decision. Taft does not preclude the consideration of such evidence. Considering evidence outside the administrative record in this circumstance will not lead to the “anomalous” result discussed in Taft. Thus, such evidence may be considered to determine if a plan administrator’s decision was affected by its conflict of interest. Reviewing that evidence, and the evidence in the administrative record, we now consider whether Tremain presented sufficient evidence to establish a rebutta-ble presumption that MetLife’s conflict of interest as insurer and plan administrator affected its decision to deny her benefits.

We confronted a similar issue in Lang . There, the plaintiff received disability payments because of anxiety, depression and insomnia. After two years her insurer terminated her disability benefits because the plan covered only a two-year period of disability due to mental disorder. The plaintiff then claimed she was suffering from fibromyalgia, a physical condition that can cause anxiety, depression and insomnia. The insurer refused to reinstate her disability benefits, claiming there was insufficient evidence that she suffered, from fibromyalgia. After receiving additional medical evidence from the plaintiff that established she did in fact suffer from fibromyalgia, the administrator again denied her disability benefits, this time on the ground that the plaintiff had not sufficiently demonstrated that (1) the fibro-myalgia alone was disabling, and (2) she was not suffering from a mental disorder in view of the fact that the symptoms of fibromyalgia were anxiety, depression and insomnia. We held that the administrator’s inconsistent reasons for denying benefits constituted material, probative evidence that its conflict of interest as plan administrator and insurer affected its benefits decision. See Lang, 125 F.3d at 799 ; Brown v. Blue Cross and Blue Shield of Alabama, Inc., 898 F.2d 1556, 1569 (11th Cir.1990) (holding that the insurer’s initial denial of coverage for two separate hospitalizations, and subsequent decision to grant benefits for the first hospitalization without any additional information, was sufficient material and probative evidence). Because the administrator in Lang failed to rebut the presumption that its benefits decision was affected by self-interest, our review of its denial of benefits was de novo. See Lang, 125 F.3d at 799 .

As in Lang , MetLife’s explanations for terminating Tremain’s benefits are troubling. MetLife denied using the Bell & Howell Plan in making its benefits decision, claiming it had only inadvertently relied on that plan’s disability definition. Yet on several occasions, and as early as 1992, MetLife requested that examining physicians use the Bell & Howell Plan’s definition to determine if Tremain was totally disabled. Further, the record presents persuasive evidence that Dr. Petrie’s conclusion that Tremain was not disabled was based on the improper Bell & Howell disability definition. In addition, MetLife informed Tremain that upon an “internal review” it had found that her earning capacity had not decreased by fifty percent. Yet MetLife failed to state what it determined to be Tremain’s earning capacity, and unlike the January 31, 1996 letter, it failed to list any occupations and corresponding salaries suitable for Tremain. There was also evidence that Tremain had suffered at least a 50% loss of earning capacity as a result of her physical condition.

We conclude that Tremain presented sufficient material and probative evidence to create a rebuttable presumption that MetLife’s decision to terminate her disability benefits was affected by its conflict of interest. MetLife has not presented any evidence to rebut that presumption. Therefore, MetLife’s decision to terminate Tremain’s benefits is reviewable de novo. See Lang, 125 F.3d at 799 .

*978 Under a de novo standard of review, the district court must review de novo the plan administrator’s decision to deny benefits. In conducting that review, the district court may decide the ease by summary judgment. It may not do so, however, if there are genuine issues of material fact in dispute. Tremain contends she raised genuine issues of material fact regarding whether she is (1) totally disabled under the Bell Plan and (2) a “salesman” for purposes of calculating her benefits. We agree.

According to the Bell Plan, after the first 24 months of disability payments, Tremain would be considered “Totally Disabled” if she either (1) was “completely and continuously unable to perform the duties of any gainful work or service for which [she is] reasonably qualified taking into consideration [her] training, education, experience, and past earnings” or (2) suffered a “50% loss of earning capacity” as a result of her illness.

Dr. Perlroth and Dr. Margóles each concluded that Tremain was totally disabled, while Dr. Petrie concluded she was not. As to her decreased earning capacity, Tre-main presented evidence that she earned over one-hundred thousand dollars in 1989, the year before she became disabled; the vocational assessment showed that Tre-main could function in numerous employment positions with median annual earnings of up to $49,072, which was less than fifty percent of her 1989 wages. The vocational assessment further indicated that some sales people could earn up to $250,-000 per year, however, the assessment did not indicate whether this was realistic considering Tremain’s condition.

We conclude that there are genuine issues of material fact as to whether Tre-main falls within either definition of total disability under the Bell Plan. The amount of Tremain’s benefits is also in dispute. The amount of those benefits will depend, in part, on whether she was a “salesman.” MetLife contends Tremain was not a “salesman” because her job title was “General Sales Manager” and when it asked Bell Industries whether Tremain was a “salesman,” Bell Industries said she was not. In contrast, Tremain presented evidence that she received commissions as a General Sales Manager and that those commissions were dependent on her sales record. Furthermore, the Bell Plan does not indicate that a “General Sales Manager” cannot also be a “salesperson” for the purposes of receiving benefits. We conclude that there is a genuine issue of material fact as to whether Tremain was a “salesman.”

Because there are genuine issues of material fact in dispute as to whether Tremain was disabled and as to whether she was a “salesman,” her entitlement to benefits and the amount of those benefits may not be decided by summary judgment. A trial is necessary. See Kearney v. Standard Ins. Co., 175 F.3d 1084, 1094-95 (1999) (en banc), cert. denied, — U.S. -, 120 S.Ct. 398 , — L.Ed.2d - (1999). In that trial, as the district court reviews de novo MetLife’s benefits decision, it will have to consider whether to admit evidence outside of the administrative record. In making that decision, the court is to be guided by its sound discretion and our decision in Mongeluzo v. Baxter Travenol Long Term Disability Benefit Plan, 46 F.3d 938 (9th Cir.1995). There, we held that a district court should admit evidence outside the administrative record when that evidence “is necessary to conduct an adequate de novo review of the benefit decision.” Id. at 944 (quoting Quesinberry v. Life Ins. Co. of North America, 987 F.2d 1017 , 1025 (4th Cir.1993)). Evidence that meets this standard need not satisfy the strict rules for the admissibility of evidence in a civil trial, and may be considered so long as it is relevant, probative, and bears a satisfactory indicia of reliability. See Mongeluzo, 46 F.3d at 941 , 943 n. 2 (doctor’s report and two affidavits admissible).

*979 CONCLUSION

In determining whether a plan administrator’s conflict of interest affected its decision to deny benefits, evidence outside the administrative record may be considered. The evidence in this case created a rebuttable presumption that MetLife’s conflict of interest affected its decision to deny Tremain benefits. MetLife failed to rebut that presumption. Accordingly, the administrator’s decision to deny benefits is subject to de novo review by the district court. Because there are genuine issues of material fact in dispute pertaining to Tremairis entitlement to benefits, a trial is necessary. In resolving the disputed factual issues at trial, the district court may consider evidence outside the administrative record “necessary to conduct an adequate de novo review of the [plan administrator’s] benefit decision.” Id. at 944 . The evidence the court may consider in conducting its de novo review need not be admissible according to the strict rules for the admissibility of evidence in a civil trial, but may be considered by the district court so long as that evidence is relevant, probative, and bears a satisfactory indicia of reliability.

REVERSED and REMANDED.

. Although Tremain did not appeal the district court's dismissal of her ERISA claims against Bell Industries and MetLife, we refer to MetLife throughout the opinion because MetLife was the Bell Plan administrator and, as a result, decided whether the Bell Plan would provide benefits to Tremain. Thus, we do not refer to MetLife as a defendant, but rather as the plan administrator.

. The Bell Plan defines "Total Disability” as when a participant:

The Bell & Howell Plan, however, defines "Total Disability” as where the beneficiary:

. The Bell Plan grants disability benefits equal to 60% of a participant's Basic Monthly Earnings, which is defined as follows: