Guide

Do programs with many IMGs sponsor visas?

293 Internal Medicine programs have a roster where over half the residents are IMGs, by ACGME's own reporting. Only 35 of those 293 also state, on their own page, that they sponsor no visa. If a high IMG share were a reliable visa signal, you would expect most of those 293 to line up with the no-visa group. They do not: the large majority of IMG-heavy programs sit outside it, sponsoring something. A roster count and a visa policy are two different facts, counted from two different sources, and they do not move together the way it is tempting to assume.

Why IMG share is not a visa signal

ACGME's IMG figure counts every resident who graduated from a medical school outside the United States or Canada, regardless of citizenship. That includes US citizens who studied medicine abroad and returned to train here, who need no visa sponsorship at all, alongside international citizens who trained abroad and do need it. A program's roster can be more than half IMG almost entirely on the strength of US-citizen graduates of Caribbean or other international schools, and such a program has no particular reason to sponsor anything, because most of its IMG residents were never going to need sponsorship in the first place. A high IMG share tells you the program hires internationally trained doctors. It does not tell you whether those doctors needed a visa to take the job.

What the 35-of-293 gap actually shows

If IMG share and visa policy were the same fact wearing two names, the 293 IMG-heavy programs and the 104 no-visa programs would mostly overlap. Instead only 35 programs appear in both groups: a large majority of the no-visa group is not especially IMG-heavy, and a large majority of the IMG-heavy group is not in the no-visa group at all. Whatever is driving a program's decision to sponsor or not sponsor a visa, it is not simply how many of its current residents happen to be IMGs. Treating a high roster percentage as a green light, or a red flag, for your own visa needs is reading a fact for something it was never measuring.

A concrete way this plays out

Picture two programs, unnamed, each reporting an IMG share above 50 percent to ACGME. The first built that roster mostly from US citizens who trained at international medical schools and returned home to match; none of them needed a visa, so the program has never had to develop a sponsorship posture one way or the other, and its own page may say nothing about visas at all. The second built its roster mostly from international citizens who do need sponsorship, and that program has necessarily worked out, and usually published, a real answer: which visa it sponsors, or that it sponsors none. Both programs would report the same IMG percentage to ACGME. Only one of them tells you anything useful about your own visa situation, and the roster number alone cannot tell you which program you are looking at.

What to read instead

The only reliable signal for whether a program will sponsor your visa is the program's own sentence about visas, quoted and dated, which is exactly the fact this site's check reads directly rather than inferring from a roster count. A roster percentage is public ACGME data about who currently trains there; a visa sentence is the program stating its own policy toward applicants who need sponsorship. They answer different questions, and only one of them answers yours. If a program's page states nothing about visas at all, that silence is itself worth noting rather than filling in from the roster: an unread or unpublished visa policy is not the same as a no-visa policy, and reading roster share as a substitute risks manufacturing a conclusion the program never actually stated. When we cannot find a program's own visa sentence, we mark it unread rather than guessing from anything else on the page, including its roster composition, because a guess dressed up as a fact is the exact failure this whole site exists to avoid.

Counted on 10 September 2026 from 654 programs read. See how we read programs for the method, and start your own list, free for the 2027 cycle.