Kirsten Majeski was employed by Metropolitan Life Insurance Company (“Met-Life”) and participated in MetLife’s Short Term Disability Plan, which is governed by the Employee Retirement Income Security Act (“ERISA”). This appeal concerns MetLife’s decision to reject Majeski’s claim for short-term disability benefits. MetLife determined that Majeski had failed to submit enough evidence to support her claim. Majeski filed suit, but the district court granted summary judgment against her. Although MetLife’s determination is entitled to deferential review, we conclude that there are such significant gaps in the evidence supporting its deci *480 sion that further proceedings are necessary.
I
Majeski worked for MetLife as a nurse consultant, which required her to sit at a desk and use a computer and tеlephone throughout the normal eight-hour workday. In June 2006, after complaining of pain and numbness in her shoulders, arms, and hands, Majeski was diagnosed with cervical radiculitis, a disorder of the spinal nerve roots. See Stedman’s Medical Dictionary 1622 (8th ed.2006). She applied for benefits from MetLife’s Short Term Disability Plan, which defines a participant as “disabled” when, as thе result of “illness or accidental injury,” she is “receiving appropriate care and treatment from a doctor on a continuing basis” and “unable to earn more than 80% of [her] pre-disability earnings at [her] own occupation for any employer in [the] local economy.” The plan grants discretionary authority to the plan administrator to intеrpret its terms and determine a participant’s entitlement to benefits. MetLife initially approved a temporary award of short-term disability benefits to allow Majeski to pursue treatment, but eventually it determined that she was not eligible for benefits beyond August 25, 2006, because, in its view, her medical records did not objectively establish any functional impairments that would prevent her from continuing her work as a nurse consultant.
Majeski appealed. In response to Met-Life’s assertion that she had not presented objective evidence establishing any functional impairments, she submitted newly obtained medical evidence. David Weiss, a physiatrist (that is, a rehabilitation specialist), completed a five-page Cervical Spine Residual Functional Capacity Questionnaire that documented Majeski’s “significant limitations” in repetitive reaching, handling, and fingering. Dr. Weiss indicated that Majeski could use her hands to grasp, turn, and twist objects for 25 percent of the workday, that she could use her fingers for fine manipulation 100 percent of the time, and that she could nоt use her arms for reaching. Dr. Weiss also reported that Majeski could not sit in a “competitive work situation” any longer than 45 minutes without needing to take a break. But in another part of the questionnaire, Dr. Weiss reported that Majeski did not have significant limitation of motion. Majeski later explained to MetLife that Dr. Weiss had misinterpreted the part of thе questionnaire where he was asked to document Majeski’s limitations in repetitive reaching, handling, and fingering. Dr. Weiss amended the questionnaire simply by crossing out “100 percent” under the column “Fingers: Fine Manipulation” and writing instead “0 percent,” indicating that Majeski could not use her fingers for fine manipulation at all.
In addition, Susan Hardin, a physical therapist, examined Majeski, tested her functional capabilities, and then submitted a Functional Capacity Evaluation Summary that documented her findings. Hardin concluded that Majeski’s limitations on sitting and typing made it impossible for her to return to her job as a nurse consultant. Hardin’s conclusion was based on a Physical Work Performance Evaluation, which consists of 36 tasks, including a 30-minute “sitting tеst.” The evaluation revealed that, although Majeski was capable of performing physical work at the medium level of exertion, she could sit only occasionally and could not type more than eight-and-a-half minutes without experiencing significant pain. (In other words, in Hardin’s view, although Majeski was able to perform at the greater exertional level of “medium,” she could not — perhaps unlike most people — handle a more seden *481 tary position.) Hardin also observed that Majeski’s cervical spine, shoulders, wrists, and elbows were capable of a range of motion within functional limits.
MetLife then asked Phillip Marion, an independent physician consultant who is board-certified in physical medicine, rehabilitation, and pain management, to review Majeski’s medical records and evaluate whether she had any functional limitations that would preclude sedentary work, particularly sitting and using a telephone and computer. Dr. Marion responded on March 1, 2007, that there were “minimal objective findings on physical and neurological examination” to support a finding of functional limitations. He added that Majeski was “otherwise independent with activities of daily living, ambulation, and not restricted from driving a motor vehicle.” Although Dr. Marion acknowledged Hardin’s finding that Majeski could perform medium-level work, he did not address either the limitations Hardin had identified on Majeski’s ability to sit and type or Hardin’s сonclusion that Majeski could not work as a nurse consultant. Nor did Dr. Marion mention Dr. Weiss’s questionnaire, which is not listed among the medical records MetLife submitted to Dr. Marion. Dr. Marion issued a second report on March 27 in which he concluded that additional medical evidence submitted by Majeski’s neurologist did not change his opinion.
On March 28, MetLife forwarded Dr. Mаrion’s reports to Dr. Weiss and asked him to respond with comments by April 10. MetLife also alerted Majeski’s counsel to the deadline. Dr. Weiss responded unhelpfully on April 6 with a single sentence: “I disagree with the decision of Dr. Marion.” On April 12, after MetLife’s deadline for comment on Dr. Marion’s reports had passed, Majeski’s counsel faxed a letter to MetLife seeking to introduce deposition testimony that Dr. Marion had recently given in an unrelated case; allegedly this testimony revealed Dr. Marion’s predisposition to rule in favor of employers and against claimants, regardless of the evidence. Because the deposition testimony ran more than 200 pages, the attachment did not accompany the fax but was instead contained on a CD that Majeski’s counsel mailed that same day. But without reviewing (and possibly before receiving) Dr. Marion’s deposition testimony, MetLife determined on April 18 that Majeski was not disabled. MetLife cited Dr. Marion’s conclusion that Majeski’s medical records neither contained objective findings nor supported an inferеnce of functional impairments.
Under the terms of MetLife’s plan, Majeski’s disability benefits could be reduced by the amount of Social Security disability benefits she was eligible to receive, whether or not she actually applied for those benefits. Majeski accordingly submitted an application to the Social Security Administration in May 2007 and received a favorable determination in March 2008.
Majeski sued MetLife in federal court, challenging the denial of disability benefits under ERISA. See 29 U.S.C. § 1132(a)(1)(B). After the parties agreed that a magistrate judge could handle the case, the district court granted summary judgment against Majeski. Because Met-Life’s plan grants discretionary authority to the plan administrator, the district cоurt ruled that it would review MetLife’s determination under the arbitrary-and-caprieious standard. In so doing, it rejected Majeski’s argument that
Metropolitan Life Insurance Co. v. Glenn,
— U.S. —,
II
A
Majeski begins with an argument that we have rejected:
Glenn,
she urges, requires a reviewing court to apply a heightened standard of review whenever a plan administrator is, like MetLife, also the payor of benefits. See
Black v. Long Term Disability Ins.,
What this court is still pondering is just
how
to consider a plan administrator’s conflict of interest. There are two possible ways to read
Glenn.
See
Marrs v. Motorola, Inc.,
But
Marrs
expressed discomfort with a standard of decision “in which unweighted factors mysteriously are weighed” and instead adopted a “more directive” reading of
Glenn
that focuses on the “gravity” of a plan administrator’s conflict of interest.
Mam,
B
Majeski next argues that, in light of Glenn and two cases from the Fifth Circuit and the Eighth Circuit that apparently endorse more searching review in conflict cases, the district court should *483 have parted ways with this court’s precedent. In conducting this review, she continues, the district court should have considered evidence that was not part of the administrative record, namely Dr. Marion’s deposition and her Social Security award.
But Majeski’s expansive reading of
Glenn
loses sight of the distinction between deferential review and
de novo
consideration. Majeski rightly observes that
Glenn
gave more weight to the plan administrator’s conflict of interest because the plan administrator there had first encouraged the claimant to file for Social Security benefits, then received the bulk of those benefits, and finally ignored the Social Security Administration’s finding when determining whether the claimant was disabled under the terms of the plan.
Nor are we persuaded by the Fifth Circuit case Majeski cites.
Vega v. National Life Insurance Services, Inc.,
Ill
That said, it remains true that deferential review is not a euphemism for a rubber-stamp. We find it troubling that Dr. Marion’s report — the sole basis for MetLife’s determination — concludes, erroneously, that Majeski did not submit objective evidence of functional limitations. Dr. Marion does not acknowledge, much less analyze, the significant evidence of functional limitations that Majeski offered. Dr. Marion notes Hardin’s conclusion that Majeski could perform medium-level work, but he ignores Hardin’s critical qualification that Majeski was neverthеless incapable of typing and sitting. Dr. Marion’s statement that Hardin’s evaluation “does not document, nor is it reasonable to conclude from it, that the claimant has functional limitations that precluded sedentary work activity requiring sitting, using a computer and telephone” is simply not true. Hardin explicitly says that Majeski cannot sit or type sufficiently to return to her former job as a nurse consultant. And Dr. Marion does not even mention Dr. Weiss’s questionnaire (nor is it listed under the documents sent to him for review).
*484
In our view, these omissions make Majeski’s case like two other recent decisions in which we have found a plan administrator’s determination arbitrary and capricious. In
Leger,
we held that it was arbitrary and capricious for а plan administrator to “ignore” and “dismiss out of hand” evidence in a functional-capacity evaluation that a claimant was not capable of sitting, concluding this was an “absence of reasoning in the record.”
We cannot square MetLife’s treatment of Hardin’s evaluation and Dr. Weiss’s questionnaire with
Leger
and
Love’s
insistence that procedural reasonableness is the cornerstone of the arbitrary-and-capricious inquiry.
Leger
explains that arbitrary-and-capricious review turns on whether the plan administrator communicated “specific reasons” for its determination to the claimant, whether the plan administrator afforded the claimant “an opportunity for full and fair review,” and “whether there is an absence of reasoning to support the plan administrator’s determination.”
We recognize that at some point we аre dealing with a question of degree. A plan administrator need not delve into medical evidence that is irrelevant to its primary concern. Nor must plan administrators annotate every paragraph of a thousand-page medical record. Closer to the line, there may be circumstances in which it would not be unreasonable if a plаn administrator inadvertently overlooked one of several medical reports that reached the same conclusion it had already rejected. But a plan administrator’s procedures are not reasonable if its determination ignores, without explanation, substantial evidence that the claimant has submitted that addresses what the plan itself has defined as the ultimate issue — here, whether Majeski’s functional limitations were objectively documented. See 29 C.F.R. § 2560.503-l(g)(iii) (requiring plan administrator to describe in adverse benefit determination “additional material or information necessary for the claimant to perfect the claim” and explain why).
Majeski has asked us to rule directly in her favor, but we are not inclined to short-circuit the process established by MetLife’s plan. When a plan administrator fails to provide adequate reasoning for its determination, our typical remedy is to remand to the plan administrator for further findings or explanations. See
Love,
Because there will be further proceedings, we address briefly Majeski’s remaining arguments, which we find to be
*485
without merit. Majeski argues that it was arbitrary and capricious for MetLife to terminate her benefits without showing that her condition had improved, but that is merely one factor to consider. See
Leger,
The decision of the district court is Vacated and the case is Remanded so that the district court may return this matter to MetLife for further proceedings consistent with this opinion.