For friends, mentors and writing groups reading a GRFP draft
NSF GRFP Review Rubric for Drafts
A rubric for reading someone's GRFP draft: NSF's review criteria word for word, evidence columns, a skim test, a rules check and a comment sheet.
What NSF asks for
The GRFP review criteria, part by part
NSF's personal statement template3 and research plan template4 list qualities reviewers "will look for evidence of." They also tell applicants they "do not need to address every bullet." Each row pairs NSF's words with what to look for as you read.
| Part | What NSF says | What to look for in the draft |
|---|---|---|
| Intellectual Merit | "The potential to advance knowledge." Factors range from the ability to "Plan and conduct research" and "Persist and overcome challenges" to grades, coursework and awards. | What the applicant did, in the first person: the question, the method and a result, with numbers where they exist. A paragraph of "we" with no "I" hides who did the work. |
| Broader Impacts | "The potential to benefit society." Reviewers may weigh "the likelihood that the applicant's activities will" reach outcomes such as "Increase participation in STEM," one of 10 on NSF's list. | Past activities with the applicant's role, who they reached and how many, then future plans that grow out of them. Look for one or two done well, not all 10. |
| Personal statement | Evidence of qualities such as "Sustained interest in STEM," "Personal initiative and motivation" and a "Vision for their future role in STEM." | Experiences that each show what the applicant learned or did next, leading to clear goals for graduate school. A CV in paragraph form shows activity, not growth. |
| Research plan | "The research idea and proposed approach," the "Resources needed," a rationale that is "clear and compelling" and "familiarity with relevant literature." | A question or hypothesis in the first paragraph, methods one student could carry out with the resources named, and a fallback if the main approach fails. |
| Both statements | "Develop a consistent theme in both statements," say NSF's tips for applicants5. | A plan that builds on experience in the personal statement, and goals and Broader Impacts plans that agree across the two. Quote both sides of any mismatch. |
| Clarity | Reviewers usually know the general field but "probably will not be experts in your specific proposed research topic." | Terms a scientist from a neighboring field wouldn't know, each with a plainer word, and results described only as "promising." |
| Format rules | Exact "Intellectual Merit" and "Broader Impacts" headings in both statements, and 3 and 2 pages counting references. Approved fonts, 6 lines per inch at most, 1-inch blank margins, alt text on figures, and no URLs or DOIs. | Check the final PDFs, not a printout. NSF's templates warn that an application breaking the format "will be returned without review." The format checker tests PDFs against these rules. |
| Eligibility | No support for research whose goals are "directly human disease- or health-related," applied crop or food safety work, policy advocacy, or conservation and resource management practice. The solicitation lists a few limited exceptions. | Goal sentences that name a disease, a treatment or a policy to change. The eligibility checker tests a topic against NSF's rules. |
Quotes are from NSF 26-5261, NSF's tips for applicants and its two statement templates, all read on October 29, 2026.
How it works
How do you review someone's GRFP draft?
To review a GRFP draft, read it against NSF's two criteria and write down the evidence for each. Intellectual Merit covers "the potential to advance knowledge," and Broader Impacts "the potential to benefit society." NSF 26-5261 says "each criterion is necessary but neither, by itself, is sufficient," so score them separately. Under each, quote the lines that show the factors NSF lists, and where nothing does, say what's missing. Leave the rewriting to the applicant, who certifies both statements are "their own original work."
Read it the way a panelist would. NSF's information for reviewers2 says each reviewer starts with about 20 applications and reads the statements beside transcripts and two or three letters. It also reminds them that GRFP "supports individuals, not projects," so read for the person's potential. Begin with a skim test: read only first sentences and bold text, then say what came through. Read each statement once without marking it, and quote evidence on the second pass. Rate each criterion last, then hand back the three fixes that matter most.
Paste a statement, or one section of it, into the box at the top and click Edit with AI. River fills the evidence columns with quotes and lists first sentences for the skim test, leaving the ratings and fixes to the reader. For River's own read, with ratings and five ranked fixes, try the mock review. To see what strong evidence looks like, read these winning GRFP examples. The 2026–27 deadlines have passed, so check the rubric against NSF's next solicitation before next year's drafts.
How it works
From a draft to three clear fixes
Download it or open it in River
Download the rubric as a Word file to print, or click Edit with AI for your own copy in River.
Paste a section
Optional. River quotes it into the evidence columns, lists first sentences for the skim test and leaves the ratings blank.
Read and rate
The reader does the skim test, reads each statement through, checks the evidence and rates each criterion last.
Hand back three fixes
Write the three most important fixes, each naming the statement, paragraph and change, or ask River to draft them.
| NSF factor | Evidence found | What's missing |
|---|---|---|
| Plan and conduct research | "I designed a 10-week survey of 40 tide pools" (para. 1) | Who analyzed the data? |
| Work independently | None found | One step taken without being asked |
| Interpret and communicate research | "I presented a poster on the survey at our campus symposium" (para. 3) | What the poster concluded |
| Persist and overcome challenges | "When the first temperature loggers flooded, I rebuilt the housings" (para. 1) | What the rerun showed |
An invented draft. River fills the evidence with quotes, and the ratings stay with the reader.
What you'll get
Score a draft against NSF's exact criteria, with quotes as the evidence
- NSF's nine Intellectual Merit factors and ten Broader Impacts outcomes, verbatim
- Evidence and what's-missing columns, a rating line for each criterion, and a skim test
- A rules check, a one-story check and a sheet for the three biggest fixes
Before and after
A statement in. Evidence with quotes out.
Research plan, first paragraph: "Last summer I designed a 10-week survey of 40 tide pools to test whether heat waves shift snail grazing. When the first temperature loggers flooded, I rebuilt the housings and reran two weeks of sampling."
Plan and conduct research: "I designed a 10-week survey of 40 tide pools" (Research plan, paragraph 1). Missing: who analyzed the data, and what it showed. Persist and overcome challenges: "When the first temperature loggers flooded, I rebuilt the housings" (Research plan, paragraph 1). Missing: what the rerun showed. Work independently: None found. Rating, Intellectual Merit: left for your reader.
Quotes only, never a rewrite. Ratings and fixes stay blank unless you ask River to suggest them.
What lands in your doc
What's inside the GRFP review rubric
- Intellectual Merit, verbatim
NSF's nine factors, each with evidence and what's-missing columns, plus a four-point check of the research plan. - Broader Impacts, verbatim
NSF's 10 outcomes laid out the same way, plus checks that past work and future plans are specific. - A rating line for each criterion
Excellent, Very Good, Good, Fair or Poor, with one sentence on why. - The skim test
First sentences and bold text only, then what the reader learned and what the full read added. - Rules, one-story and clarity checks
Headings, page limits, fonts, margins, references, alt text and eligibility, then whether both statements agree, and jargon. - A comment sheet
A rating summary, what's working, and the three most important fixes, each tied to a paragraph.
Questions GRFP applicants ask
Common questions
What rubric do NSF GRFP reviewers use?
GRFP reviewers use NSF's two merit review criteria, Intellectual Merit and Broader Impacts, giving "balanced consideration to all components of the application." The solicitation lists factors under each, and NSF's personal statement3 and research plan4 templates list what reviewers "will look for evidence of." This rubric quotes the solicitation's lists. Its Excellent-to-Poor ratings are for your readers, not NSF's form or a forecast.
What are the Intellectual Merit criteria for the GRFP?
Intellectual Merit covers "the potential to advance knowledge." Reviewers may consider the applicant's ability to plan and conduct research, work independently and as part of a team, and interpret and communicate research. NSF also lists taking initiative, thinking creatively, solving problems, and persisting through challenges. Grades, coursework and awards count as evidence of intellectual ability.
What counts as Broader Impacts for the GRFP?
Broader Impacts covers your potential to benefit society through your activities. NSF lists 10 example outcomes, including increasing participation in STEM, improving STEM education, public engagement with STEM, well-being and national security. You don't need all 10. A strong section shows past activities with your own role, and plans that grow out of them.
Who should review a GRFP draft?
Two kinds of reader, if you can find them: someone in your field to judge the science, and someone from a neighboring field to catch jargon. Labmates, mentors, writing center tutors and writing groups all work. NSF's tips for applicants5 say to leave time to "get feedback from others," so share drafts weeks before the deadline, not days.
Can I use this rubric for the next GRFP cycle?
Yes, after one check. The rubric quotes NSF 26-526, the solicitation for the 2026–27 cycle, posted August 25, 2026. Its application deadlines ran from October 19 to 23, 2026, and NSF expects to announce results by early April 2027. When NSF posts the next solicitation, compare its criteria and format rules with the rubric before your group starts reading.
Is this GRFP review rubric free?
Yes. Download the rubric as one Word document, with no account or signup, and print it or fill it in on screen. Edit with AI opens the same rubric in River, which is free to start. There you can fill it in, share it or ask River to change it. Paid plans add more AI use and access to the most advanced models.
What does Edit with AI do?
It opens the rubric in your own River workspace with NSF's criteria in place. If you paste a statement first, River fills the evidence columns with short quotes, lists first sentences for the skim test, and checks the headings, references and eligibility. Ratings and fixes stay blank unless you ask River to suggest them, and it never rewrites the statement.
Your next step