A request for evidence is rarely asking you to prove more. It is telling you the file did not prove what you said it did.

The language of a request for evidence, read as a translation. Each row is a request form that recurs across extraordinary ability, exceptional ability and national interest cases.
What the notice requests	What the officer is actually saying
Evidence that the contributions are of major significance to the field	The exhibits show the work exists. Nothing in the record shows the field noticed, used or built on it.
Clarification of the field of endeavour claimed	As described, the field has no ascertainable top, so the central question cannot be answered either way.
Evidence of the significance of the person's leading or critical role	The letters state a job title. They never state what would have failed without this person in it.
Documentation that published material was not created for the purpose of marketing	The media exhibit is a press release with the applicant's name in it.
Evidence that the criteria have been satisfied over a sustained period	Every qualifying exhibit is dated inside the six months before filing.
Updated letters providing specific information on the achievements claimed	The letters restate the regulation and then restate the CV, in the third person, at length.
Evidence of the relevant comparison group for the claimed high remuneration	The salary is high compared with people who do different work in a different market.

The short version

A request for evidence is not the beginning of a denial. It is a written statement that the record, as it stands, does not establish something the law requires. Those are different events. The first is a question with a deadline attached. The second is a conclusion. Most practitioners who read requests for evidence as a category rather than as individual events end up with the same working view: the great majority of them are not about whether the person is good, but about whether the file proved it.

That distinction matters because it changes what you do about it. If an RFE were a judgment on merit, the response would be to argue harder. Because it is usually a judgment on the record, the response is to close a specific evidentiary gap — and, more usefully, the same gap can be found and closed before anything is filed at all.

The mechanism is unglamorous. The regulation places the burden of establishing eligibility on the applicant, at the time of filing. The officer then decides the case on the evidence of record. Where the evidence of record is judged insufficient, the agency may issue a request for evidence rather than deny — which is a form of procedural generosity, not a signal that the case is weak.

So the useful question is never "why is the officer asking for this?" It is "what in the file made this question available?" The answer is almost always one of four things: an exhibit that is responsive but not probative, a significance claim made without significance evidence, a field defined so loosely that no comparison group exists, or a record compressed into a window too narrow to demonstrate a career.

Everything below is an expansion of those four. The section after them is about letters, because letters are where the largest single gap usually sits — they are the evidence applicants spend the most time on and the evidence officers most often find deficient.

What an RFE is usually telling you

The exhibit answers the criterion and proves nothing

This is the most common pattern by a wide margin. The petition asserts a criterion, attaches an exhibit that is technically on the subject, and the exhibit does not demonstrate the thing asserted. Judging the work of others is claimed with evidence of winning a competition. Authorship is claimed with a conference abstract. A leading or critical role is claimed with an organisation chart. Media recognition is claimed with coverage of the employer. Each exhibit is responsive to the criterion's subject matter and none of it is probative of the criterion's requirement. USCIS states the operative test directly: the criterion is met where the evidence objectively establishes the described achievement, not merely where documents touching on the topic are attached.

The significance is asserted, not established

A second, subtler pattern: the fact is documented and the interpretation is missing. There is a citation count and no account of what the cited work changed. There is a deployed product and no demonstration that the field treats it as significant. There is a patent and nothing showing anyone wanted it. This is where the manual is unusually concrete. It points to published research that has provoked widespread commentary on its importance from others in the field, and to documentation that work has been highly cited relative to others' work in that field, as material that may be probative of significance. The direction is the point: significance is shown by what other people in the field did in response, not by what the applicant says the work means.

The field has no ascertainable top

The central question in an extraordinary ability case is whether the person is among the small percentage at the very top of a field. That question cannot be answered if the field is defined so broadly that nobody is at the top of it, or so narrowly that the comparison group is a team of four. A claimed field that shifts between the petition, the letters and the exhibit list guarantees a request for clarification, because the officer has no fixed population against which to measure the claim. Where the standards do not readily apply to an occupation, comparable evidence is available — but a general assertion that a criterion does not apply is not probative, and a claim of comparability has to be argued on the specifics, not asserted as a formality.

The record was built on a deadline

This is the request that should genuinely change behaviour, because it cannot be argued away. Instead of disputing any exhibit, the notice observes that the qualifying evidence clusters into a short period immediately before filing. That observation bears directly on whether acclaim is sustained, and sustainable acclaim is a maintained condition rather than a historic one — the manual directs officers to consider whether the person has held a comparable level of acclaim to the one originally recognised, or whether it has since lapsed. A file assembled in one push to satisfy a checklist reads as exactly that. There is no response to it except a longer chronology, and a longer chronology does not exist on the deadline.

Letters that describe rather than attest

What the manual says letters are for

USCIS addresses support letters directly, and the passage is worth reading twice. Many petitions contain them; they are not without weight; but they should not form the cornerstone of a successful claim. The statements made by the witnesses should be corroborated by documentary evidence in the record. The letters should explain in specific terms why the witnesses believe the person to be of the calibre claimed. Letters that merely reiterate the agency's own definitions, or that make general and expansive statements about the person and the work, carry little weight.

Description is not attestation

Most deficient letters are not dishonest. They are descriptive. They recount the applicant's history, restate the significance of the project, list the publications, and conclude that the person is outstanding. A describing letter tells the officer what the writer knows. An attesting letter tells the officer what the writer is in a position to know that nobody else is. The difference is specificity of vantage: this person's method is the reason our trial succeeded; this person's patent is licensed by three competitors; I have reviewed the work of two hundred people in this field and this is the only case in which a technique from this subfield was adopted across all of them. None of that can be written by someone who has only read the CV.

Corroboration is the real requirement

The manual's instruction to corroborate is where most letter-based responses fail. A letter asserting that a contribution was adopted by the field is an assertion; it becomes evidence when the record also contains the independent artefact of adoption — the licensing agreement, the citation by a competing group, the third party who built on the method, the standard that incorporated it. Send letters alone and the officer has, in substance, two unverified statements: the applicant's and the writer's. The remediation is structural rather than rhetorical. For each fact that only a witness can supply, there should be a document in the record that would be different if the fact were untrue.

The officer is asking for the thing the record should already have shown

Read a request for evidence as a diff. It is a list of requirements minus the evidence, and the officer has done the subtraction for you. That is an uncomfortable document to receive, but it is an unusually informative one, because it is the closest thing to a marking scheme that an adjudication produces.

The reason the subtraction exists comes from how adjudication actually works. Officers handle a wide range of classifications and are instructed to evaluate the record as a whole, considering all evidence in the totality, weighing some items more heavily on their own and others more persuasively in combination. They are not specialists in the applicant's subfield. That is the operative fact. If the importance of a contribution is visible only to the twelve people who work in that specific area, it has not been established for the audience that decides the case — not because the audience is unsophisticated, but because the petition's job was to translate, and the translation is missing.

This is also why requests for clarification of the field are so damaging and so common. An undefined field removes the comparison group, and without a comparison group "at the very top" becomes a phrase rather than a finding. An officer cannot conclude that a person is among the small percentage at the top of a field if the record does not fix which population the person is being placed in and at what rank.

And it is why quality is assessed where a criterion contains a qualitative requirement. Meeting the description of a criterion is step one. Whether the evidence is of a calibre that supports the claim is a question the officer is instructed to engage with rather than defer. A file built to clear step one, with no regard for how each item reads on its own merits, is a file that has optimised for the wrong stage.

What an RFE is not

It is not a denial, and treating it as one produces bad responses. Panic produces length. Length without precision is the failure mode: a two-hundred-page submission that restates the original petition, adds more of the same exhibit type, and answers the officer's specific question obliquely, if at all. The officer asked one thing. Answering it is more persuasive than burying it.

It is not a licence to file a new case inside the old one. A response is constrained by the criterion the officer is asking about, and by the record that already exists. Evidence that goes to a requirement nobody raised does not help, and evidence that changes the underlying theory of the case invites the question of whether the original filing was ever accurate.

It is not, on the other hand, an opportunity you should under-use. The record after a response is a complete record, and an officer decides on that. Where the missing corroboration genuinely exists — the adoption evidence, the third-party validation, the specific letter from someone with direct vantage — the response is the moment to put it in. The cases that go wrong after an RFE are usually not under-evidenced. They are over-argued and under-documented.

One practical mechanic to plan around: in premium processing a request for evidence stops the clock, and a new processing period begins only when the response is received. The published service periods are 15, 30 or 45 days depending on form and classification. If you paid for speed, an RFE is the event that hands the schedule back to the agency.

How to read one as a diagnosis

First, count the requests. A single narrow request for clarification is a substantially different event from four requests that each go to a different criterion. The second pattern usually means the officer does not accept the shape of the case, and at that point the question is whether the theory of the filing can survive being answered in writing.

Second, classify each request as either a documentation gap or a translation gap. Documentation gaps are solvable: the corroborating artefact exists and was not submitted. Translation gaps are harder, because they mean the record contains everything and the significance argument was never made. Respond to the first with documents and to the second with an explanation grounded in documents — never with the same letter tightened up.

Third, check every response against the exhibit it answers. If the officer asked for evidence of significance and the response is a letter asserting significance, the gap has been restated rather than closed. The test is mechanical: what in this response would be different if the claim were false?

Fourth, and the point of the whole exercise: treat the list as a pre-filing audit for the next case. The four patterns above are not random. They are the predictable result of building a record against a checklist rather than against a standard, and they are all detectable before filing by anyone willing to read the file as an officer will — as a stranger who needs to be told, in writing, why it matters. The best way to answer a request for evidence is to make it unnecessary.

Questions this raises

Is a request for evidence the same as a denial?

No. Where the evidence of record is judged insufficient, the agency may issue a request for evidence rather than deny, and then decide the case on the completed record. It is a statement that the file does not yet establish a requirement — not a finding that the requirement cannot be met. Many cases with RFEs are approved.

Does an RFE stop premium processing?

Yes. Where a petition requires additional evidence or a response to a notice of intent to deny, the premium processing period stops and resets, and a new period begins when the agency receives the response. The published periods run 15, 30 or 45 days depending on the form and classification involved.

Should I respond to an RFE or withdraw and refile?

Respond when the gap is documentation. Refile when the request shows the underlying theory was not accepted — where the field itself is disputed, or four separate criteria are each questioned. Withdrawing and refiling can preserve the chance to rebuild the case properly, at the cost of the original filing date and the fees.

Can I submit evidence in an RFE response that was not in the original petition?

Yes. The decision will be made on the evidence of record as it stands after the response, and evidence supplied in response becomes part of that record. What an RFE response should not do is change the theory of the case, because that invites the question of whether the original filing was accurate.

Why do letters of support sometimes fail to satisfy an RFE?

Because the agency directs that letters should not be the cornerstone of a claim and that their statements should be corroborated by documentary evidence in the record. Letters that merely repeat the agency's own definitions, or make general and expansive statements about the person, carry little weight however senior the writer.