Federal courts reviewing an ERISA benefits denial usually read only what the insurer had during the internal appeal, which makes the appeal file the entire case.

Most ERISA benefit cases are decided on written submissions rather than witnesses in a courtroom. A treating physician who would have been compelling on the stand only counts if their opinion is in the file.
The denial letter said the claimant could perform sedentary work, cited a file review by a physician who never examined her, and closed with a sentence about the right to appeal within 180 days. She appealed on her own, sent a two-page letter and a note from her rheumatologist, and lost again. By the time an attorney read the file, the second denial had issued and the internal appeal was exhausted. The medical evidence that would have carried the claim, three years of infusion records and a functional capacity evaluation, existed. It was simply never sent to the insurer.
In most federal circuits, a court reviewing a denied claim under the Employee Retirement Income Security Act does not hold a trial in any recognizable sense. There is no jury, no live testimony from the treating physician, and usually no discovery beyond narrow questions about conflict of interest. The judge reads the administrative record: the documents the plan administrator had in front of it when it made the final decision. Evidence created afterward, however persuasive, generally sits outside that record and cannot be considered. That single procedural fact reorganizes the entire strategy of a disability, health, or pension appeal.
The practical consequence is blunt. A claimant who wins at the appeal stage wins because the file was complete; a claimant who loses in court often loses because the file was thin and the closing came before anyone realized it. The Department of Labor is responsible for the claims procedure regulations that set the deadlines and the disclosure duties on both sides, and those rules are what give a claimant the leverage to build the record properly while the leverage still exists.
A careful reader treats the denial letter as an index rather than a verdict. It should list every document the reviewer relied on, and that list is the first thing to compare against reality. If the insurer names four office visits and the treating chart contains nineteen, the reviewer decided the case on a fifth of the evidence. If a file-review physician is quoted, the report itself should be requested, in full, along with the reviewer's specialty, the questions posed, and any surveillance, internal claim notes, or vocational analysis in the file. Under the claims regulations, that material is available on request at no charge, and asking for it is routine rather than adversarial.
What emerges from that comparison is usually a specific gap. The insurer applied an occupational definition drawn from a job description the claimant has never seen. Or it credited an examination that lasted twenty minutes over eight years of treatment. Naming the gap precisely is what turns an appeal letter into an argument.
Treatment records prove diagnosis. They rarely prove function, which is the question a disability plan actually asks. A chart that says the patient reports severe fatigue is weaker than a functional capacity evaluation measuring sitting tolerance in minutes, lifting in pounds, and reliability across a two-day protocol. An attending physician statement that tracks the plan's own language, addressing whether the claimant can perform the material duties of her own occupation, does more work than a supportive letter written in general terms. Neuropsychological testing, cardiopulmonary exercise testing, and pain journals kept contemporaneously all convert a subjective complaint into something a reviewer must engage with in writing.
Vocational evidence matters just as much and is skipped more often. If the plan defines the occupation as it is performed in the national economy, a vocational expert can establish what that occupation genuinely requires, and whether the restrictions in the medical file are compatible with it. Wage data and occupational classifications are tracked federally, which gives that analysis a foundation the insurer has to answer.
Before sending, a careful reader checks four things: that every treating provider's complete chart is enclosed, not summarized; that the submission responds by name to each reason the denial gave; that a proof-of-delivery method establishes the date; and that anything still pending, a test scheduled for next month or a report not yet returned, is flagged in writing with a request to hold the record open. That last request is small and frequently granted.
The appeal stage feels like paperwork and functions like a trial. Treating it that way, with the evidence gathered before the deadline rather than after, is what puts a claimant in front of a judge with something to read.