Radiologist shortage 2026: why 47% of postings are re-posts

Four of five full time diagnostic radiologists walked out of the Washington DC VA Medical Center inside six months. The chief of radiology was one of them. By July 2025 the department had stopped doing body MRI, by that September a CT backlog had formed, and studies were going out to community imaging centers because there was nobody left inside the building to read them. The VA Office of Inspector General published its account on August 5, 2026, and named the cause plainly: a return to office mandate that spooked a department into leaving.

Six weeks earlier, a job market analysis had argued that the national radiologist shortage is largely a counting error.

Both things are true, and holding them together is the whole story. If you run imaging at a hospital or an imaging center, the number that governs your year is not the national supply of radiologists. It's whether the specific people who can cover your specific hours will move to your specific town. That question has a different answer in Omaha than it does in Tampa, and the gap between those answers is where most coverage plans quietly fail.

A disclosure, because it shapes what follows. AstraRad's 240 subspecialists read for more than 160 imaging centers and radiology groups, none of which solved their coverage by getting a radiologist to move house. We are a party to this argument. Saying so here also makes the frame explicit: everything below treats location and work arrangement as the variables, because that is the part of the problem remote reading touches. The hiring math is real too, and it is the part we cannot help with.

What did the VA inspector general actually find?

The OIG reviewed radiology staffing and services at the DC facility after departures surfaced in the press. It confirmed the exodus, traced it to the January 20, 2025 presidential memo requiring full time in person work for federal employees, and described that policy as the tipping point, while noting it was not the sole cause.

The operational damage followed a familiar sequence. First a modality goes dark, in this case body MRI in July 2025. Then a backlog forms in the highest volume study type, CT, by September. Then the work leaves the building, to community imaging centers that bill more than in house staff would have cost. The OIG flagged six patient cases as carrying particular risk of harm. It also found that facility leaders had not notified quality management of known problems and had not tracked incomplete studies, which is the part that turns a staffing gap into a safety finding.

One detail deserves more attention than it got. The facility was granted a one year remote work exemption, but with no clarity on whether it would be extended. That ambiguity did its own damage: you cannot recruit a radiologist onto a work arrangement you cannot promise will exist in thirteen months. The OIG noted that teleradiology and remote work have become factors in employment decisions for many radiologists, influencing recruitment and retention strategies across the industry. That sentence, in a government report about one hospital, is the most portable finding in the document.

Is there really a national radiologist shortage?

Depends what you mean by national. The 2026 RadBoard.io radiology job market report, published June 8, 2026 from a study window of March 9 to May 26, looked at 20,775 job postings and found only 10,979 unique roles behind them. About 47% of open jobs were re-posts of the same position, with the average role appearing 1.89 times across 18 different sources.

That's a meaningful correction. If you have been reading that there are more than 1,400 open radiology positions and drawing conclusions about supply, roughly half of that number is the same job counted twice. Postings are a measure of advertising, and advertising is cheap.

But the report does not say the shortage is fictional. It says it has a shape. Nebraska and Minnesota were carrying more than 50 positions open past two months. Florida and Texas filled fastest. About one in five postings, some 1,470 roles, stayed unfilled beyond two months, which is a long time to run a department a body short. The analysis put it bluntly: the shortage isn't national, it is concentrated in places radiologists won't move to.

Subspecialty matters as much as state. Roles filling inside 14 days ran at 79.9% for pediatric radiology and 78% for emergency radiology, against 27.2% for musculoskeletal and 29.3% for breast imaging. If your open role is a breast imager in a rural market, the national averages are telling you almost nothing about your search.

The practice model mattered too. Private equity backed platforms filled 99% of their 988 openings within 60 days. Academic systems filled 55% of 165. Some of that is compensation and some is recruiting infrastructure, and the report does not fully separate them.

Where the two findings meet

Here is the reconciliation, and it is worth sitting with because most coverage of this week's news picked one side and ran with it. The RadBoard analysis measures the market for radiologists who are willing to relocate for a job, because that is what a job posting is: an offer contingent on somebody moving to where the scanner is. Under that constraint, supply looks tight in Nebraska and loose in Florida, exactly as reported. The VA finding measures something adjacent and more revealing. Those DC radiologists did not leave because there were no radiologists. They left because the terms changed, and the terms in question were about where they had to sit while reading. Four of five chose a different employer over a different commute, in a market where the alternatives were plentiful enough to make that choice easy. Put the two together and the constraint stops being headcount and becomes willingness: how many radiologists will accept your hours, your location and your work arrangement, at your rate. Every facility that has run a nine month search for a night reader already knows this in their bones. The two reports simply put numbers underneath it, from opposite directions, in the same month.

What the posting data cannot tell you

Two cautions before anyone forwards the 47% figure to a board as proof that hiring is easier than they thought.

The first is that a re-post is ambiguous. A role listed 1.89 times across 18 sources may be one recruiter spraying the same vacancy across every board, which is the duplication the report corrects for. It may also be a role that was filled, then reopened three months later when the hire did not stay. The second case is a worse signal than a straightforward vacancy, and posting counts cannot separate the two. Attrition running at 2.5% a year suggests some share of those repeats are the second kind.

The second caution is that job postings only count demand somebody decided to advertise. A department that gave up on recruiting two years ago and now runs permanent locum coverage generates no posting at all. Neither does a group that absorbed a departure by pushing call onto the remaining partners. Both are unmet demand, and neither appears in a scrape of 18 job boards. The measured market is the market that still believes a search will work.

These two cautions push in opposite directions, which is roughly why the honest read is the one the report itself gives. There are fewer distinct open roles than the raw counts imply, and the roles that are open are concentrated where they are hardest to fill. Neither correction changes what a director in Nebraska does on Monday morning.

What this means for a facility covering nights

The practical translation is unglamorous. If your coverage gap is in a market that fills fast, in a subspecialty that fills fast, a search is a reasonable plan and you should run one. If either of those is false, a search is a hope, and hope has a burn rate. One in five postings running past two months means the median unsuccessful search costs you a full quarter of degraded coverage before you learn it failed.

The ACR Health Policy Institute's work force update supplies the trend line underneath all of it. Radiologist attrition rose from 1.1% a year in 2014 to 2.5% by 2022, having already reached 2.0% before the pandemic, so this is not purely a COVID artifact. Over the same period the number of practices employing radiologists fell 14.7% while the work force itself grew 17.3%, pushing average practice size from 9.7 to 17.9 radiologists. Consolidation is absorbing the supply. The projection puts radiology at roughly 90% adequacy in 2038.

Ten percent short, for twelve years, is not a crisis you solve. It's a condition you staff around.

Which coverage models survive a bad search

Three approaches carry a facility through an open search, and each trades something different.

Locum coverage buys a warm body on your premises. It is the most expensive per shift, has the longest lead time, and stops the moment the invoice does. Useful for a known, bounded gap. We have written a fuller comparison of locum radiologist versus teleradiology economics if you are pricing both.

Redistributing call internally is free in cash and expensive in everything else. It works for a few weeks. Past that you are spending the goodwill of the people you most need to keep, in a market where 2.5% annual attrition means somebody is already being recruited. The five levers that actually move radiologist burnout are worth reading before choosing this one for a long gap.

Contracted remote reads convert a fixed hiring problem into a variable per study cost. The lead time is days instead of months, the coverage does not take vacation, and the arrangement scales down as cleanly as it scales up. What it does not do is put a colleague in your reading room, sit on your committees, or handle procedures. If your gap is a body of work instead of a body, that trade is fine. If you need a person in the building, it isn't, and you should run the search.

There is a fourth honest option that rarely gets written down: reduce the service. Stopping body MRI, as the DC facility effectively did, is a decision. It is better made deliberately, with a plan and a date, than discovered in September when the backlog surfaces.

The question worth asking your own numbers

Before choosing, get two figures on paper. First, how many hours a week are currently uncovered or covered by somebody working past a reasonable shift. Second, what your last search cost in elapsed months. If the second number is over three and the first is over ten, the search is not your coverage plan, whatever the org chart says.

Then decide what you are actually buying. A facility in a fast filling market with a generalist vacancy should recruit. A rural critical access hospital with one overnight CT a night has never been able to justify a hire and should stop treating that as a failure, which is the case we make in more detail for rural and critical access coverage. Most facilities sit between those poles and end up running a search and a coverage contract at the same time, which is the correct answer more often than people expect.

How AstraRad fits

We read and sign final reports for imaging centers, hospitals and radiology groups, billed per report, with no minimums. The panel is 240 board certified subspecialists across ten subspecialties, reading on scheduled US shifts, and every measured figure we publish sits dated on our SLA and QA page. STAT is under one hour by contract with a measured median of 30 minutes, and 99.4% of reports came back inside their tier over the trailing 12 months.

What that buys a facility in the position this article describes is time. Coverage starts in days, so a search can run at its own pace instead of at the pace of an uncovered Tuesday night. If the search succeeds, the contract scales down. If it doesn't, nothing breaks in September.

To see what contracted coverage of your volume would look like, request a volume quote and ask for the licensure and capacity answers up front. If you are earlier than that and still mapping the options, the comparison of in house versus outsourced radiology lays out the cost structures side by side, and the rest of the buying research lives in the teleradiology resource library.