Of the ten criteria in 8 CFR 204.5(h)(3), the fourth is the one most people have already met and fewest people have documented. Almost every researcher who has published has been asked to review. Many practitioners have sat on award panels, grant committees or conference programme committees. And when the file is assembled, this criterion is often the thinnest in it.
The reason is a single distinction that the Policy Manual states directly.
The invitation is not the evidence
USCIS puts it plainly: the petitioner must show that the person has not only been invited to judge the work of others, but also that the person actually participated in the judging.
An invitation e-mail proves you were asked. It proves nothing about whether you did it. The guidance names what closes the gap — a request from the journal to do the review, accompanied by proof that the review was actually completed.
So the unit of evidence here is a pair, not a document:
- the request, showing you were selected to evaluate someone else's work, and
- confirmation that the review was delivered.
Files that contain only the first half invite an RFE on precisely that point, and the person receiving it usually has the second half somewhere — they simply never thought to include it.
Where the proof actually lives
For peer review, the editorial platforms keep it. Editorial Manager, ScholarOne and OpenReview all maintain a reviewer history that can be exported, and journals will generally confirm completed reviews when asked. Some publishers issue reviewer certificates. Services that aggregate verified review records exist for the same reason.
For panels, committees and juries, the proof is usually the appointment letter plus something showing the work happened: the programme listing you as a judge, minutes, the scoring sheets, a thank-you from the organiser naming what you evaluated.
For grant review, the agency confirmation of panel service is typically enough, and it is usually retrievable years later.
What does not satisfy it
Supervising your own team. Evaluating the work of people who report to you is management, not judging the work of others in the field. The criterion contemplates peers outside your own line of authority.
Being on an editorial board without reviewing. The title is not the activity. A board membership with no completed reviews proves the appointment and not the judging — and may belong under the membership criterion instead.
One review. A single completed review technically shows participation, and it is thin. The second step of the analysis asks whether the whole record shows sustained acclaim, and one review in a career answers that question badly.
Why this criterion is worth getting right
It is one of the three that most records can genuinely reach without doing anything new — the work was already done. Unlike original contributions, which needs evidence of adoption by the field, or press, which depends on journalists deciding to write about you, judging depends on records you can retrieve.
That makes it the cheapest criterion to document properly, and the most wasteful one to lose on a missing confirmation e-mail.
Where to look next
The criterion sits inside the two-step analysis described on the EB-1A page: three of ten, and then the final merits determination on the whole record. The criterion that fails most often for the opposite reason — evidence that proves originality but not significance — is original contributions of major significance. To see which criteria your own record already reaches, the assessment reads them one by one.