Graduate stipends in U.S. biomedical science sat near 25,000 dollars or below for years before rising to 26,352 dollars in fiscal 2023 on the NIH's standard predoctoral pay line, with a further increase following for 2024. Because that stipend is the salary of a person in their late twenties often living in an expensive city, researchers and policy analysts argue it functions as a filter on who can afford to enter science at all, one that favors the financially secure.
This site publishes information about research training policy, not financial advice.
What do graduate students actually earn?
The reference point is the National Research Service Award, or NRSA, pay table that the NIH publishes each fiscal year and that many institutions mirror. The entry predoctoral stipend was 26,352 dollars for fiscal 2023 after years of stagnation around 25,000 dollars or below, and the NIH announced an additional increase for fiscal 2024 following sustained pressure from trainee groups. Postdoctoral stipends on the same tables started in the high fifties. Universities pay from a mix of mechanisms: NIH training grants tied to NRSA levels, individual fellowships, research-grant salary lines and teaching assistantships, which is why two students in the same department can earn noticeably different amounts.
Relative to living costs, these figures are tight. Surveys by trainee unions and professional societies have repeatedly found sizable fractions of doctoral students reporting food or housing insecurity, findings consistent with studies of graduate-student well-being published through the late 2010s and 2020s.
How can a stipend filter who enters science?
The mechanism is arithmetic rather than mysterious. A five-to-six-year PhD at roughly 26,000 to 38,000 dollars a year, typical ranges at large biomedical programs in recent years, means delayed savings, delayed home ownership and reliance on family money for emergencies. Students whose families cannot buffer that gap face a choice peers with support do not. Economists describe this as ability-to-pay selection, distinct from ability selection, and it compounds at every step: unpaid research assistantships as undergraduates, moving costs for campus visits, low-paid postdocs afterward.
Unionization grew as a response. Graduate-worker unions expanded notably from 2020 onward, including multi-campus organizing within the University of California system in 2022, where strike actions won pay increases, and at private universities after the NLRB's evolving stance on graduate-employee bargaining.
What do the data show about diversity effects?
Direct causal evidence is limited, but multiple observational strands point the same way. NIH's own internal analyses, published in the 2010s and early 2020s, found that Black and Hispanic applicants for research grants were funded at lower rates than white applicants and that early-career gaps compound later. Surveys of doctoral students, including the federal Survey of Earned Doctorates, show first-generation and lower-income students carrying more debt at graduation. Program-level studies have found that better-funded stipends correlate with more diverse applicant pools, though these are correlational, unable to fully separate pay from program prestige or location. The honest summary is that stipend levels plausibly narrow the pipeline, with the strongest evidence coming from debt and attrition data rather than from any controlled experiment.
How do we know stipends stagnated?
The NRSA pay tables are public and complete, which makes this unusually checkable: between the early 2000s and the early 2020s the entry predoctoral stipend rose from the low twenties to roughly 25,000 dollars, a real decline once inflation is applied, before the 2023 and 2024 increases restored some ground. NIH leadership acknowledged the problem, and advocacy groups such as the Future of Research nonprofit and petitions organized in 2022 documented the gap against local rents. Limitations: institution-paid stipends vary above the NRSA floor, private universities pay more on average than public ones, and cost-of-living adjustments mean the same stipend buys very different lives in Houston versus Boston. No single national number exists, only floors and distributions.
Could raising stipends change science?
Supporters, citing the correlation above, argue higher and more uniform pay widens the applicant pool and reduces attrition, which wastes training investment, since NSF estimates of PhD completion suggest a meaningful share of entrants never finish. Skeptics, including some funders, note that stipend increases flow through to grant budgets, allowing fewer funded positions for the same pot, a real trade-off Congress weighs in appropriations. There is also a sequencing problem: raising predoctoral pay without raising postdoctoral pay compresses the ladder, making the PhD-finishing wage nearly identical to the fresh-PhD wage. What most policy analysts appear to agree on, across NIH advisory reports from the 2020s, is that predictability matters as much as level, because institutions plan cohorts years ahead.
What should a prospective student take from this?
That stipends are negotiable facts, not weather. The NRSA tables are public, unions have changed outcomes at specific campuses, and programs differ in whether they top up fellowships or cover fees. The deeper lesson is structural: the price of becoming a scientist is paid partly in foregone income during the years when peers are earning, and who can pay that price shapes the profession's demographics long before any hiring committee meets.
How do U.S. stipends compare internationally?
Comparison with Europe is sobering for the American model. Doctoral candidates in countries including Germany, the Netherlands and Switzerland are frequently employed as staff or funded through structured programs, with net incomes commonly well above U.S. predoctoral levels and access to social benefits built on employment status, arrangements documented in European science-organization salary surveys through the 2010s and 2020s. U.S. stipends instead arrive as educational awards, which affects everything from retirement contributions to unemployment eligibility, a distinction graduate-union campaigns have made central. The trade-off historically cited for the American system, shorter time-to-degree and higher lab resources, has narrowed in the data as European programs professionalized, according to doctoral-education studies published in the 2020s.
For more context, read Why foreign-born scientists matter to U.S. research.
For more context, read community colleges stem.
For more context, read What research says about learning loss recovery.