UNUM Life Insurance Company of America appeals from the order of the district court granting Wanda Stup summary judgment on her claim that UNUM improperly denied her long-term disability benefits in violation of the Employee Retirement Income Security Act of 1974 (ERISA). UNUM does not dispute that Stup suffers from lupus, fibromyalgia, and other afflictions. Rather, UNUM argues that it acted reasonably in determining that, despite these disabilities, Stup could perform a job commensurate with her training and, therefore, was not entitled to long-term disability benefits under the ERISA plan. The district court rejected this argument and concluded that UNUM abused its discretion in denying Stup long-term benefits. For the reasons that follow, we affirm.

Lupus is a "chronic inflammatory connective tissue disorder" that can affect the joints, skin, kidneys, lungs, heart, and blood. See The Merck Manual 426-29 (17th ed.1999); see also U.S. Dept. of Health and Human Services and Nat'l Institutes of Health, Systemic Lupus Erythematosus 3 (rev.Aug.2003), http:// www.niams.nih.gov/hi/ topics/lupus/slehandout/slehandout.pdf. Symptoms, which may come and go, include extreme fatigue and malaise, arthritis, unexplained fever, chest pain, pneumonia, and skin rashes. U.S. Dept. of Health and Human Services, supra, at 3, 8-10; The Merck Manual, supra, at 426-29. Doctors use tests that detect the level of specific antibodies in the blood and skin biopsies to diagnose lupus. U.S. Dept. of Health and Human Services, supra, at 11-13.

Fibromyalgia is a rheumatic disease with similar symptoms, including "significant pain and fatigue," tenderness, stiffness of joints, and disturbed sleep. Nat'l Institutes of Health, Questions & Answers About Fibromyalgia 1 (rev. June 2004), http:// www.niams.nih.gov/hi/topics/fibromyalgia/Fibromyalgia.pdf. See also Ellis v. Metro. Life Ins. Co., 126 F.3d 228, 231 n. 1 (4th Cir.1997) (quoting Taber's Cyclopedic Medical Dictionary (16th ed.1989)); Sarchet v. Chater, 78 F.3d 305, 306-07 (7th Cir.1996). Doctors diagnose fibromyalgia based on tenderness of at least eleven of eighteen standard trigger points on the body. Sarchet, 78 F.3d at 306. "People with rheumatoid arthritis and other autoimmune diseases, such as lupus, are particularly likely to develop fibromyalgia." Nat'l Institutes of Health, supra, at 4. Fibromyalgia "can interfere with a person's ability to carry on daily activities." Id. at 1. "Some people may have such a severe case of fibromyalgia as to be totally disabled from working, but most do not." Sarchet, 78 F.3d at 307 (citations omitted).

In the mid-1990s, Stup was diagnosed with lupus and fibromyalgia.1 The record reveals that a Board-certified rheumatologist, Dr. Nathan Wei, began treating Stup in January 1996 and monitored her closely for the next five years, examining Stup bi-weekly or monthly.

A biopsy conducted in January 1994 verified that Stup suffered from lupus erythematosus. Periodic lab tests consistently showed high levels of an antibody associated with lupus in Stup's blood. Stup's medical records evidence that she has suffered the following manifestations of lupus: rashes, lesions, livedo, swelling, temperature fluctuations, chest pains, and pneumonia.

Stup's fibromyalgia is also well documented. She initially responded well to the treatment prescribed by Dr. Wei, but her condition rapidly worsened. Dr. Wei's notes from his frequent examinations of Stup state that she had diffuse trigger point tenderness and complained of fatigue, weakness, and pain. Dr. Wei changed Stup's medications regularly to try to alleviate her pain, often to no avail.

Dr. Wei recommended that Stup, along with taking prescribed medications, "limit her activities" and "take daily naps as necessary." But Stup disregarded his advice by continuing to work as an administrative assistant to the president of Frederick Underwriters, Inc., where she had been employed since 1984. In Dr. Wei's view, Stup "attempted to work far past her physical abilities" for several years. By March 16, 1998, Stup's debilitating pain and fatigue prevented her from performing her duties. Only then did she apply for disability benefits under Frederick's ERISA plan.

Frederick's plan is contained in an UNUM insurance policy, which UNUM administers. The policy entitles a "Class 2" employee (UNUM's classification of Stup) to twenty-four months of disability benefits if the insured cannot "perform each of the material duties of h[er] regular occupation" (emphasis added). Under the policy, disability benefits for Class 2 employees cease after two years unless "the insured cannot perform each of the material duties of any gainful occupation for which [s]he is reasonably fitted by training, education or experience" (emphasis added). UNUM paid Stup disability benefits from September 1998 until November 2000, at which time it terminated her benefits, explaining that "[s]ince the definition of disability now applies to `any occupation' we feel you could return to work in a sedentary occupation."

Stup appealed UNUM's denial of long-term disability benefits. In support of her appeal, she submitted her voluminous medical records and an UNUM Estimated Functional Abilities Form filled out by Dr. Wei, stating she could lift only one to ten pounds occasionally and was "unable to perform even sedentary [sic] for more than 1-2 hours at a time."

UNUM referred her medical file to Dr. Parke, an in-house physician, who determined that "[b]ased on medical (objective) studies, Ms. Stup has evidence of a [Connective Tissue Disorder]." He concluded that the "best way to fully assess" her ability to work "would be a visual assessment, possibly coordinated" with a physical and rheumatological independent medical examination. Instead of following this recommendation, UNUM's associate medical director, Dr. Bielawski, decided to obtain a Functional Capacity Evaluation (FCE). Accordingly, UNUM requested that Dr. Wei write Stup a prescription for an FCE, which he did.

On April 3, 2001, a physical therapist conducted the FCE. The therapist reported that Stup's flexibility and range of motion were within functional limits but that she "[e]xhibited pain behaviors during testing." The therapist could not measure Stup's strength because Stup "g[a]ve out" before an adequate assessment could be made. Similarly, the therapist could not determine Stup's aerobic capacity because Stup could not move "at the minimum speed necessary for the treadmill walking test (2.0 mph)." Stup was able to walk "for 4.5 minutes at a speed of 1.2 mph prior to stopping the test due to hip and leg pain"; she averaged about 7.5 pounds of grip strength with her right hand, and 11.1 pounds with her left. Tests also showed that she could occasionally lift 16.5 pounds from floor to knuckle, 12.5 pounds from knuckle to shoulder, and 7.5 pounds from shoulder to overhead and could carry 12.5 pounds for 100 feet. Stup performed inconsistently on squat lift, isometric pull, and hand grip tests, which the therapist interpreted as a sign that Stup did not make a consistent effort. The FCE did not measure Stup's positional tolerance because after the remaining tests were explained to her, Stup "indicated she did not feel she could go on" and the physical therapist "agreed that testing should not continue."

In total, the Functional Capacity Evaluation lasted only two and a half hours. The physical therapist opined that "[t]he results of this evaluation indicate" that Stup could perform sedentary work, but immediately cautioned that "the results of this evaluation may not be truly indicative of the clients [sic] functional capabilities." The therapist noted "inconsistencies" that "became apparent in the examination." For example, Stup showed stronger grip strength in her left hand than her right even though she is right handed and the dominant hand is usually ten to fifteen percent stronger than the other. The therapist also pointed out that at the time Stup complained of severe pain, her heart rate was lower than it had been after thirty minutes of sitting. The physical therapist concluded "it would not be prudent to make recommendations regarding specific job duties that this client can or cannot perform due to a lack of consistent and true information."

Nevertheless, an UNUM doctor, Burton McDaniel, after stating that he reviewed Stup's file "with special attention to recent FCE," concluded that the FCE "appears to be very thorough and valid" and the results "concur with the statement that she is capable of performing sedentary level work." His one-paragraph report focused entirely on the FCE test results. Using as a baseline the measurement of Stup's heart rate during a stress test taken the previous year, which was referred to in the FCE, the doctor implied Stup had not given a full effort during the FCE because if she had, her heart rate would have been higher during testing. He concluded that [w]ithout inconsistencies claimant may well be able to function at a higher work level such as light." UNUM then informed Stup that its initial decision "to deny [her] claim was appropriate."

Stup asked UNUM to review its decision a second time, challenging UNUM's assertion that she could perform sedentary work.2 In support of her appeal, Stup submitted an affidavit in which she explained her daily activities:

On an average day, I don't go anywhere. I go outside in the morning to get to the mailbox, have breakfast and then read or nap, depending on my pain level. I try to stay awake in the morning and sometimes I will make efforts to pick up around the house. My daughter helps me clean up, and I try to do a little bit each day. Some days I cannot do any cleaning or picking up because of the pain; other days I can do a few minutes. I cannot do any type of "normal" house chores at a "normal" pace. I only live in one half of my house because there is no way that I could take care of a house. Sometimes I eat a light lunch and sometimes I skip lunch. Then I take my pain medication and lay down for the afternoon. I watch the news, and on some days I go sit outside on the front porch and read. I eat dinner sometimes. I then watch TV or sit on the porch.

Stup also stated in her affidavit that she had offered to try to continue the tests for the FCE, but the physical therapist "said that she'd rather I not start the tests if I wasn't going to complete them, and that she didn't think I was going to make it."

In addition, Stup provided UNUM with a detailed letter from Dr. Wei dated August 20, 2001, explaining his diagnosis and opinion that Stup could not perform even sedentary work:

Ms. Stup's fibromyalgia diagnosis is supported by the following criteria: positive reaction to all trigger points, nine positive tender points, non-restorative sleep, short term memory deficiency, aches and pains with minimal activity, and severe, persistent fatigue on minimal exertion. Ms. Stup's fatigue is best described as profound and prevents her from being able to perform normal daily activities, such as personal grooming and household chores, without experiencing draining, marked loss of energy. Her short-term memory deficiency is evident in her inability to remember simple directions and the purpose for which she sets out on driving excursions.

In addition to fibromyalgia, Ms. Stup suffers from low back pain that radiates down her legs as a result of degenerative disc disease. As for her lupus erythematosus, the most significant symptom is limited mobility and pain in both her hands.

I have recommended to Ms. Stup that she limit her activities, take daily naps as necessary to replenish her energy level, and continue with medications and follow-up treatment to document and monitor her symptoms. Despite my recommendations, she attempted to work far past her physical abilities. I have explained to her that she will experience symptom flare-ups and periodic improvement relative to her baseline condition. I also explained that the ability to predict when she may experience flare-ups or periods of improvement is minimal at best.

Based upon my history of treatment with Ms. Stup, I do not believe that she is physically capable of performing even sedentary work. She must be able to perform physical activities at her own pace with frequent rest periods; she needs to nap on a daily basis; and she will suffer an exacerbation of symptoms as a result of stress common in any workplace. Furthermore, as a result of fatigue and back pain, she is limited in her ability to stand, sit or walk for periods of time greater than thirty (30) minutes. I recommend against repetitive bending, kneeling, crawling, and climbing stairs. She is also limited in her ability to lift, carry, push or pull more than ten (10) pounds. The limitation of the use of her hands makes anything more than simple grasping beyond her ability....

Ms. Stup has never shown herself to be a malingerer or less than fully motivated to do whatever is necessary to improve her health. She is compliant with all recommendations for treatment, except to the extent that she worked past the time that it was advisable for her to do so. It is my professional opinion that Ms. Stup is totally disabled from all work. This total disability from all work has been present since March 16, 1998. Due to the chronic nature of her condition, [sic] will be totally disabled to work into the foreseeable future.

An in-house UNUM doctor, Maureen Lee, reviewed Dr. Wei's letter to determine whether it should affect UNUM's "prior opinion that claimant has sedentary work capacity." On a brief form, the doctor stated that because it was not accompanied by new objective testing, "[t]he updated medical letter does not change the prior conclusions to the file (based on FCE 4-4-01)." UNUM again denied Stup's disability claim.

Having exhausted her administrative appeals, Stup filed this suit, claiming UNUM violated ERISA by wrongfully denying her long-term disability benefits. After consideration of memoranda submitted by the parties, the district court granted summary judgment to Stup. UNUM appeals.

We review the grant of summary judgment de novo, "employing the same standards applied by the district court" in reviewing the fiduciary's decision to deny ERISA benefits. See Sheppard & Enoch Pratt Hosp., Inc. v. Travelers Ins. Co., 32 F.3d 120, 123 (4th Cir.1994) (citation omitted). Two facts affect which standard we apply in reviewing UNUM's decision to deny ERISA benefits.

First, the ERISA plan (Frederick Underwriters' insurance policy with UNUM) provided UNUM with "the discretionary authority both to determine an employee's eligibility for benefits and to construe the terms of this policy." Although this grant of authority is not as broad as that in some ERISA plans, see, e.g., Booth v. Wal-Mart Stores, Inc., 201 F.3d 335, 343 (4th Cir.2000), it does invest UNUM with discretion. When an ERISA plan affords an administrator discretion, a court reviews the administrator's decision to deny benefits for an abuse of that discretion, asking whether the denial of benefits was reasonable, Bernstein v. CapitalCare, Inc., 70 F.3d 783, 787 (4th Cir.1995) (citation omitted), "based on the facts known to [the administrator] at the time."3 Sheppard & Enoch Pratt Hosp., 32 F.3d at 125. An administrator's decision is reasonable "if it is the result of a deliberate, principled reasoning process and if it is supported by substantial evidence." Bernstein, 70 F.3d at 788 (internal quotation marks and citation omitted).

The other fact determinative of the appropriate standard of judicial review is that in denying benefits, UNUM acted under a conflict of interest. That is, its decision to deny benefits impacted its own financial interests because it "both administers the plan and pays for benefits received by its members." Id. at 788; see also Doe v. Group Hospitalization & Med. Servs., 3 F.3d 80, 86 (4th Cir.1993). As we noted in Booth, the Supreme Court has directed that this conflict "must be weighed in determining whether there is an abuse of discretion." Booth, 201 F.3d at 342 (emphasis in original) (citing Firestone Tire & Rubber Co. v. Bruch, 489 U.S. 101, 115, 109 S. Ct. 948, 103 L. Ed. 2d 80 (1989)). In these circumstances, "we will not act as deferentially as would otherwise be appropriate.... [T]he fiduciary decision will be entitled to some deference, but this deference will be lessened to the degree necessary to neutralize any untoward influence resulting from the conflict." Doe, 3 F.3d at 87; accord Ellis v. Metro. Life Ins. Co., 126 F.3d 228, 233 (4th Cir.1997); Bedrick v. Travelers Ins. Co., 93 F.3d 149, 152 (4th Cir.1996) (citing Bailey v. Blue Cross & Blue Shield, 67 F.3d 53, 56 (4th Cir.1995)). Under this sliding-scale standard of review, "[t]he more incentive for the administrator or fiduciary to benefit itself by a certain interpretation of benefit eligibility or other plan terms, the more objectively reasonable the administrator or fiduciary's decision must be and the more substantial the evidence must be to support it."4 Ellis, 126 F.3d at 233.

We cannot hold that UNUM's decision to deny benefits was either supported by substantial evidence or the product of a principled reasoning process.

Stup has offered overwhelming and uncontradicted evidence that she suffers from lupus and fibromyalgia, as well as a number of other afflictions. UNUM does not dispute or question this evidence. Rather UNUM insists Stup's ailments do not render her unable to perform sedentary work commensurate with her training.

On this issue, Stup submitted her medical records and the detailed opinion of Dr. Nathan Wei, the Board-certified rheumatologist who had treated her for years. Dr. Wei reported that Stup's fibromyalgia caused her "non-restorative sleep, short-term memory deficiency, aches and pains with minimal activity, and severe, persistent fatigue," concluding that her fatigue and memory loss prevent her from "being able to perform normal daily activities" or "remember simple directions." Dr. Wei also concluded that Stup's degenerative disc disease caused "low back pain that radiates down her legs" and that her lupus erythematosus caused "limited mobility and pain in both her hands." He said she cannot use her hands for more than simple grasping and her ability to lift, carry, push, or pull is severely limited.

Dr. Wei explained that he had urged Stup to limit her activities and "take daily naps as necessary to replenish her energy level," but that she had "attempted to work far past her physical abilities." He stated Stup "must be able to perform physical activities at her own pace with frequent rest periods," and "stress common in any work place" would "exacerbat[e]" her symptoms. Dr. Wei concluded that Stup is not "physically capable of performing even sedentary work"; rather in his "professional opinion" she is "totally disabled from all work." Stup's medical records substantiated her long history of and treatment for these disabilities.

Unquestionably, this evidence satisfies Stup's initial burden of submitting proof that she could not perform even sedentary work and was entitled to long-term benefits under the ERISA plan. Just as clearly, this evidence would not require UNUM to award benefits if it had "substantial evidence" that Stup could perform sedentary work. Bernstein, 70 F.3d at 788 (internal quotation marks and citation omitted). And as UNUM correctly notes, an administrator does not act unreasonably by denying benefits if the record contains "conflicting medical reports." Elliott v. Sara Lee Corp., 190 F.3d 601, 606 (4th Cir.1999).

UNUM maintains that it acted reasonably in denying benefits because the record contains conflicting medical evidence as to Stup's ability to work. But UNUM ignores that while an administrator does not necessarily abuse its discretion by resolving an evidentiary conflict to its advantage, the conflicting evidence on which the administrator relies in denying coverage must be "substantial" — especially when, as in this case, the administrator has an economic incentive to deny benefits.5 See Ellis, 126 F.3d at 233-34 (finding administrator with incentive to deny benefits acted reasonably in doing so because even though several doctors said insured was disabled, it had "substantial evidence" that her doctors did not agree on the proper diagnosis and three independent medical panels "concluded that there was no conclusive diagnosis"). For example, an administrator operating under a conflict of interest does not act reasonably in denying benefits if faced, on the one hand, with substantial evidence of disability and, on the other, with only tentative and ambiguous evidence that might, or might not, favor denial of benefits. This is precisely the situation at hand.

UNUM claims the FCE conflicts with Dr. Wei's assessment, insisting the FCE "unequivocally" shows that Stup can perform sedentary work. But there is nothing unequivocal about the results of Stup's FCE — indeed, the physical therapist explicitly equivocated, noting the inconsistency of the test results, in the course of ultimately offering her negative interpretation. Upon examination, it is clear that the FCE test results either supported Dr. Wei's assessment or were, at the very least, ambiguous; the results certainly do not provide substantial evidence that Stup could perform sedentary work.

First, and most obviously, the FCE lasted only two and a half hours, so the FCE test results do not necessarily indicate Stup's ability to perform sedentary work for an eight(or even four-) hour workday, five days a week. Even if the results of the FCE had shown conclusively that Stup could perform sedentary tasks for the duration of the test, which they do not, those results provide no evidence as to her abilities for a longer period. Perhaps that is why Dr. Parke — the first UNUM in-house doctor to review Stup's file and who agreed with Dr. Wei's diagnosis — thought that observing Stup as she performed her daily routine would provide the "best" method for determining her ability to work.

Second, Stup's performance on the FCE may actually substantiate rather than conflict with Stup's evidence as to her inability to work. Based on his diagnosis, the accuracy of which is not in dispute, Dr. Wei concluded that, due to pain and fatigue, Stup could not perform sedentary work for more than one or two hours at a time, had limited ability to lift or carry more than ten pounds, and could not sit or walk for more than thirty minutes at a time. Consistent with Dr. Wei's assessment, during the FCE Stup exhibited pain when her flexibility and range of motion were tested. She "gave out" before her strength could be measured. She could not walk fast enough for the physical therapist to assess her aerobic capacity, and she could only walk for four and a half minutes at 1.2 miles per hour before succumbing to hip and leg pain. The physical therapist halted the testing after only two and a half hours because it did not appear that Stup could complete the tests, suggesting Stup's exertion could not be sustained for much longer than two and a half hours. These results entirely accorded with Stup's evidence. Thus the actual results of the FCE cannot be said to conflict with Dr. Wei's conclusions and they surely do not provide substantial evidence conflicting with Dr. Wei's assessment.

In fact, the only "evidence" UNUM offered that Stup could per-form sedentary work is the physical therapist's equivocal and tentative interpretation of the FCE results and Dr. McDaniel's concurrence in that interpretation.6 But this "evidence" is neither "substantial" nor the product of the type of "deliberate, principled reasoning process" indicative of a reasonable determination. Bernstein, 70 F.3d at 788 (internal quotation marks and citation omitted).

An equivocal opinion — especially one based on ambiguous test results — simply does not provide "substantial evidence." Here, the physical therapist herself twice expressly recognized the ambiguity of the FCE results and hedged her negative interpretation of them. Initially, she warned that Stup's inconsistent test results "may not be truly indicative of the clients [sic] functional capabilities." And in conclusion, the therapist cautioned, "it would not be prudent to make recommendations regarding specific job duties that this client can or cannot perform due to a lack of consistent and true information."

Moreover, we cannot say that the reasoning of UNUM's in-house doctor, Dr. McDaniel, was "deliberate" or "principled" here.7 After reviewing the FCE, but without ever seeing or examining Stup or even acknowledging the physical therapist's warning that it would not be "prudent" to use the FCE results to determine Stup's ability to perform "specific job duties," Dr. McDaniel simply stated that the test results "concur[red] with the statement that she is capable of performing sedentary work" — ignoring that the test results could be interpreted just as easily as consistent with Dr. Wei's conclusion that Stup could not work. Thus, UNUM's doctor created conflicting "evidence" by disregarding the testgiver's own admonition as to the possible unreliability of the FCE and interpreting the results against the claimant when they could be read as entirely consistent with the medical evidence demonstrating disability.8

In sum, the only "evidence" that UNUM offered to justify its denial of long term benefits to Stup simply does not constitute "substantial evidence" arrived at after a "deliberate, principled reasoning process" — especially given that UNUM acted under a conflict of interest. See Bernstein, 70 F.3d at 788 (internal quotation marks and citation omitted). Thus, the district court did not err in finding UNUM abused its discretion in denying Stup long-term disability benefits.

Stup provided UNUM with years of substantial medical evidence supporting her diagnosis, which UNUM does not dispute, and her doctor's well-documented conclusion that she could not work. UNUM argues it acted reasonably in denying Stup benefits because it relied on evidence conflicting with her long-time physician's substantiated opinion that her medical condition significantly limited her capabilities. But the so-called conflicting evidence, upon examination, fully supports the opinion of Stup's Board-certified rheumatologist or is, at the very least, so ambiguous that the person administering the FCE thought the test results were not a prudent basis for assessing Stup's specific job capabilities. Thus the evidence on which UNUM relies is neither substantial nor conflicting.

Accordingly, the judgment of the district court is

AFFIRMED.