You have two ways to pick a fellowship. You can rank programs on interest, reputation and how bad the fellowship-year hours are, or you can start with the attending job you want five years from now and work backward to the training that gets you there. Most residents do the first. This entry in our residency survival guide makes the case for the second.
The timing makes this concrete. For the NRMP Medicine and Pediatric Specialties Match, the rank order list certification deadline was November 18 at 9 p.m. ET, and Match Day is December 2. If you certified a list this week, the decision is mostly made. If you are a PGY-1 or PGY-2 looking ahead, or you get an off-cycle offer, you still have room to run the check below.
Why the job, not the fellowship, is the real choice
A fellowship lasts two or three years. The job pattern it leads to can last decades. Yet residents compare fellowships on what the next three years feel like, and assume the job market at the end will sort itself out.
Residents on Reddit this week described where that breaks. One resident loves ICU work but wants to avoid nights for good, and has no idea how common days-only intensivist jobs are or what they pay. An intern wants to combine nephrology and critical care and can’t tell whether hiring managers want that profile or just tolerate it. A PGY-3 finally got an offer from a niche fellowship they had wanted for years, in a different state from their partner, and is now weighing an outpatient job instead. Each one is a question about the job, being asked after the training decision is underway.
Option one: rank on the training itself
This is the default, and it isn’t foolish. Case volume, procedural exposure and mentorship shape how good a clinician you become, and a strong mentor can open doors that no job board lists. The cost is that the training environment tells you little about the job market. A fellowship can be excellent at producing academic intensivists while the role you want, say a community ICU position on a week-on, week-off block, barely appears in its graduates’ placements.
Fellowship hours deserve less weight than residents give them. A resident asking how GI fellowship nights and weekends compare with IM residency is asking a fair question, but it covers three years. The call structure of the jobs GI graduates actually take will matter for far longer.
Option two: rank on the job waiting after it
Here you name the job first: the setting, schedule, region and rough pay you want. Then you check whether that job exists in volume, and which training path employers expect for it.
Sometimes the check confirms your plan. Critical care is a good example of a schedule assumption worth testing early. A quick search of current postings shows employers advertising dedicated critical care nocturnist roles and part-time intensivist positions, which tells you the shift-based model is real. Whether the mirror image, a days-only role with no nights at all, is common in the region you want is something to count, not assume.
Sometimes the check changes the length of training. A dual path costs time. The University of Pittsburgh’s combined nephrology-critical care program, for example, runs three years with at least 20 months of clinical training and 10 of scholarly work, and takes one fellow a year. That is a reasonable price if hybrid jobs exist where you want to live. It is an expensive one if they don’t, and you end up practicing as a nephrologist who keeps ICU skills you rarely use.
How to run the check in an afternoon
Write one sentence describing the job: “Days-only intensivist, community hospital, within an hour of my partner’s city.” Then do four things.
First, search current job listings for that exact role in that region, including on PhysEmp, and count them. Ten postings and two postings are different answers.
Second, ask each program you are considering where its last three to five classes went. The question that gets useful answers is: “How many of your recent graduates took jobs like this, and where?” A program that can’t answer is telling you something.
Third, find one attending who has the job you want, through a program alumnus, a society listserv or a cold email, and ask what they would train in if they started over. Ask what the job pays compared with the standard version, because a restriction such as no nights can change the pay, and you want that number before you rank, not after.
Fourth, put location on the table early. If your partner’s career fixes the city, the job market in that city is the market that matters, and a top-ranked fellowship elsewhere has to be weighed against two or three years apart or a move afterward.
When the training-first choice still wins
If you are not sure what job you want, training breadth and strong mentorship are a sound hedge, and you can narrow later. The job-first method works best when you already know a constraint you won’t trade: nights, a city, a practice model. Residents who know that constraint and still rank on prestige are the ones who end up surprised.
The plain test is this: if the job at the end of this fellowship turned out to be rare in the place you want to live, would you still choose it? If the answer is no, you have more research to do before you sign anything.