NSF Cut Required Reviews From Three to Two. The Real Story Is What Program Officers Can Now Decide Alone.

The change went live on December 8, 2025 with the kind of dry agency notice that almost no one outside the institutional research office reads carefully. NSF's Office of the Director circulated a memorandum reducing the minimum number of external reviews per full proposal from three to two, making panel review optional rather than required, permitting one of the two reviews to be conducted internally by NSF staff, and contracting panel summaries to a recommended three to five sentences. The agency framed the package as a flexibility measure designed to address proposal backlogs and the substantial workforce reduction that NSF absorbed across calendar year 2025.

Five months later, every principal investigator submitting to NSF in fiscal 2026 is operating inside a merit review system that does not work the way the proposal preparation guides still describe. The procedural changes — fewer reviews, shorter summaries, optional panels — are the surface of a much more significant institutional shift. NSF has consolidated decision authority in its program officer corps in a way that has no precedent in the post-1981 history of the agency's merit review framework, and the implications for how a competitive proposal is constructed are larger than any single procedural rule.

For background on the broader 2026 NSF funding landscape, including the $8.75 billion appropriation and the 1,752 grants terminated by DOGE earlier in the year, see our budget analysis. What follows is about the review mechanics themselves.

What actually changed in the merit review process

The pre-December rule was that every full proposal under the standard NSF merit review framework required a minimum of three external peer reviews from subject-matter experts identified by the program officer. In practice, most proposals attracted more than three — five to seven was a common range for competitive directorates — and the program officer was expected to convene a panel of those reviewers to discuss each proposal in person or by videoconference. The panel produced a summary statement, typically several paragraphs long, that synthesized the strengths and weaknesses identified across the individual reviews and assigned the proposal a rating in the agency's competitive, recommend-for-funding, do-not-fund taxonomy.

Under the December rules, the minimum drops to two reviews. The justification requirement that previously applied to proposals with only two external reviews — a written explanation by the program officer of why the panel could not find a third reviewer — has been eliminated. One of the two reviews may now come from NSF internal staff rather than an external expert, which is a substantive change because internal review by program officers has historically been treated as a triage step before external peer review, not as a substitute for it. Panel discussions are now discretionary rather than mandatory; the program officer decides whether the proposal warrants a panel based on factors that are not fully specified in the new guidance. Panel summaries, when written, are limited to three to five sentences.

The rationale offered by NSF leadership is operational. The agency lost a substantial fraction of its program officer corps in 2025, including a large number of rotators — visiting scientists who serve two-to-four-year terms as program managers and who historically have made up a meaningful share of the workforce in the science directorates. The agency also absorbed shutdown-related proposal backlogs and is operating under a fiscal year 2026 appropriation that requires it to obligate awards on a compressed timeline. The merit review changes were framed as a flexibility measure to allow the remaining program officer workforce to clear the queue.

That framing is accurate as far as it goes. It is also incomplete. The accumulated effect of the procedural changes is that a single program officer can now process a proposal with two reviews (one of which may be the officer's own), no panel discussion, and a three-sentence summary statement. The substantive decision authority that used to be distributed across a five-to-seven-person panel of external experts is now consolidated in the program officer alone for any proposal the officer chooses to process that way.

Why this is bigger than a procedural change

External peer review under the previous framework was not just a quality check. It was a mechanism for distributing decision authority across the research community in a way that constrained any individual program officer's ability to drive funding outcomes toward a personal research agenda. The five-to-seven-person panel was, in effect, a peer audit of the program officer's judgment — a check that the proposal selected for funding was one that the broader research community would also recognize as the best of the cohort.

That check is now optional. Program officers retain it for proposals they choose to send to panel, and they can opt out for proposals they do not. The agency has not published guidance on what factors should drive that choice, which means it will be driven by individual program officer judgment, subject to whatever directorate-level oversight emerges over the coming year.

The consolidation effect is amplified by two other features of the December changes. Internal review by NSF staff now counts toward the two-review minimum, which means the program officer can, in principle, write one of the two reviews themselves. The three-to-five-sentence summary statement requirement compresses the documentary record of the decision in a way that makes it harder for any subsequent reviewer — an inspector general audit, a congressional inquiry, a Freedom of Information Act request — to reconstruct the reasoning that drove the funding decision. The historical record of NSF merit review has been one of the most thoroughly documented decision processes in federal science funding. The new record will be substantially thinner.

None of this is a claim that NSF program officers will systematically misuse the new authority. The vast majority of program officers are former or current research scientists with deep subject-matter expertise and strong institutional incentives to make defensible funding decisions. But the structural shift in where decision authority sits is real, and it interacts with the broader political pressure on the agency in ways that researchers preparing proposals need to understand.

What this means for how a proposal should be written

The practical implications for proposal preparation are substantial and largely under-discussed in the institutional research office community.

First, the audience for a proposal has narrowed. Under the previous framework, a proposal had to persuade five to seven external experts plus a program officer. Under the new framework, it may only need to persuade two people, one of whom is the program officer who selected it for processing. The strategic implication is that the program officer's priorities — both the priorities published in the program solicitation and the priorities that the officer signals informally through conference presentations, listserv posts, and pre-submission inquiries — matter more than they used to. Researchers who do not contact program officers before submission are losing more competitive ground than they did a year ago.

Second, the proposal needs to stand independently in a way that previous proposals did not. Under the old framework, a panel discussion could surface strengths in a proposal that individual reviewers had missed and could correct weaknesses that one reviewer had over-emphasized. Under the new framework, that surfacing and correction mechanism is optional. A proposal that depends on panel dynamics to surface its strengths is now significantly less likely to be funded. Every claim, every figure, every section needs to be self-contained and persuasive on the page.

Third, the broader impacts section requires fresh attention. NSF's broader impacts criterion has been a moving target for two decades, and the December review changes coincided with a parallel shift in how NSF directorates are interpreting broader impacts in light of the DOGE grant terminations earlier in 2025. Proposals that include broader impacts language tied to diversity, equity, and inclusion frameworks have been disproportionately represented in the terminated grant pool. Proposals submitted in 2026 should articulate broader impacts in terms that align with the workforce, infrastructure, and national-competitiveness framing that the agency is currently emphasizing. This is not a recommendation to abandon equity-focused impact work; it is a recommendation to frame it in language that does not invite termination.

Fourth, the time-to-decision is compressing. The agency's stated goal in introducing the changes was to expedite award and obligation of FY 2026 funds, and the early evidence from the December-to-March award announcements is that decisions are coming faster. Researchers should plan their submission timing with shorter expected response windows in mind, particularly for programs that have shifted to rolling submission models.

What to do before submitting

Contact the program officer six to eight weeks before the target submission date. The program officer's view of the proposal now matters more than it did, and the only reliable way to learn what the officer is prioritizing is to ask. Researchers who treat the published solicitation as the full statement of priorities are leaving competitive ground on the table.

Complete the mandatory research security training that NSF rolled out in parallel with the merit review changes. The training requirement is a hard gate; proposals from researchers who have not completed it will not advance.

Build the proposal as a self-contained document that does not depend on panel discussion to surface strengths. Write the project summary, the broader impacts section, and the intellectual merit case as if they will be the only sections any reviewer reads carefully — because for some proposals, they will be.

Pay attention to the three-to-five-sentence summary statement constraint when reading the feedback on declined proposals. The brevity is not a signal that the proposal was dismissed without consideration; it is the new format. Researchers preparing resubmissions should treat each sentence in the summary as carrying significantly more weight than the equivalent sentence in a pre-December summary would have carried.

The longer-term question

The December merit review changes are framed as a temporary flexibility measure. NSF leadership has not committed to making them permanent, and the National Science Board has indicated that it will review the changes during fiscal 2026 with an eye toward whether to retain, modify, or reverse them.

The most likely outcome, based on the trajectory of similar institutional changes at NIH and DOE, is that the December changes will harden into the new baseline. The workforce reductions that motivated the changes are not going to reverse quickly, the proposal backlogs are not going to clear on their own, and the broader political pressure on the agency to demonstrate operational efficiency creates incentives for leadership to retain the flexibility that the new framework provides.

Researchers preparing NSF proposals for the next two to three years should plan around the assumption that the December framework is the steady state, not the exception. The merit review system that produced the agency's most-cited research portfolio is not the system that will produce the next decade's portfolio. The substantive work of adapting to that shift — in how proposals are constructed, in how program officers are engaged, in how broader impacts are framed — is the work of fiscal 2026.

The agency's stated commitment to maintaining the rigor of merit review is genuine. The structural changes to where decision authority sits are also genuine. Both are true, and the researchers who navigate the next funding cycle most successfully will be the ones who treat the consolidation of authority in the program officer corps as the operating reality, not the procedural detail.