Asked to Write Your Own Letter of Recommendation? A Guide

Asked to Write Your Own Letter of Recommendation? Now What?

"Sure, I'd be happy to recommend you — just send me a draft and I'll sign it."

If a professor, PI, or manager has just said some version of this to you, the first thing to know is that you are not in a weird edge case. A poll by the fellowship-listings site ProFellow found that 79% of respondents had been asked to prepare a draft of their own recommendation letter by a referee. Academia Stack Exchange has a canonical thread that opens: "My supervisor and two other professors… asked me to write my own recommendation letter and they will sign it" — and the answers, from working academics, mostly confirm how routine the request is.

The second thing to know is that "routine" and "safe" are not the same thing. At least one major university treats a self-drafted letter as an application violation no matter who suggested it, most portals are engineered on the assumption it does not happen, and a self-drafted letter that reads self-drafted can quietly hurt you even where no rule forbids it. Nobody official talks about any of this — the practice lives in forum threads, and the policies live in fine print. This guide connects the two.

Why recommenders ask

Understanding the request helps you answer it. The usual reasons, in descending order of comfort:

• Volume. Faculty write dozens of letters per cycle, each submitted separately to every individual program portal — a student with eight programs means eight uploads. A draft collapses an hour of recall into ten minutes of editing.
• Information asymmetry. A manager two levels up genuinely may not remember which projects you led. Your draft is, in their mind, a very thorough brag sheet.
• A test. Some mentors use it deliberately: show me how you see yourself, and I'll correct it. These recommenders usually rewrite heavily — which is the healthy version of this arrangement.
• Indifference. The writer plans to sign whatever arrives. This is the version that should worry you, ethically and strategically.

Is it actually allowed? Check before you type

There is no single rule — there is a patchwork, and you need to know which patch you are standing on.

Some universities ban it outright. Stanford's graduate application terms are the clearest in the country: "Drafting, writing, translating, or submitting your own reference, even if asked to do so by a recommender, is a violation of our application terms" — recommenders must be the sole authors, and the applicant is responsible for telling them so. Note the "even if asked" clause: your professor's request does not immunize you.

Most portals are engineered against it. Per-recommender upload links, institutional-email requirements, and verification callbacks all exist to keep the letter out of applicant hands. Several universities reserve the right to contact recommenders and verify authorship, and treat misrepresentation as grounds for denial or rescission — we documented the enforcement landscape school by school in our guide to what US universities really do about applicant-written LORs, which also covers the international-applicant context where this practice is most institutionalized.

Law school raises the stakes. LSAC's misconduct rules cover submitting "an altered, nonauthentic, or unauthorized letter," and U.S. News has warned law applicants directly that recommender-invited self-drafting still sits inside that risk zone. If you are applying through LSAC, read our law school letters of recommendation guide and be very conservative here.

The health-professions services write the rule into the platform. CASPA — the PA school application — states evaluations cannot be completed or submitted by the applicant or another party on the evaluator's behalf, on top of an applicant certification with real enforcement teeth; the same Liaison-platform structure runs dental and veterinary admissions. Medical school's version of this terrain is mapped in our self-drafted med school letter guide.

Fellowships have professional ethics codes — and attestations. The fellowship-advising profession explicitly discourages referee-invited self-drafting — ProFellow's own guidance walks through why, and through the alternatives below. GRFP and Fulbright references are confidential by design, and Knight-Hennessy requires applicants to affirm they did not write, draft, edit, translate, or submit their letters — the fellowship-specific rules and scripts are in our fellowship reference self-draft guide.

If any program on your list publishes Stanford-style language, the answer is simply no — and you now have the perfect script: "I'd love to, but [Program] makes me confirm that recommenders are the sole authors — I could send you detailed bullet points instead?"

The ethics spectrum (know where you are standing)

The practice is not one thing. It is a spectrum, and the ethical line is crossed somewhere in the middle:

1. Brag sheet / bullet points. You supply facts, dates, projects, outcomes; they write. Universally accepted — this is just being a good letter-requester.
2. Structured draft as raw material. You write a draft; the recommender substantially rewrites, recalibrates the praise to what they actually believe, and owns the result. Common; defensible in most systems (not at Stanford-style programs); the recommender's edit is what makes it their testimony.
3. Verbatim signature. You write it; they sign it unread or barely read. Now the "witness" never testified. Even where no rule names this, the letter misrepresents its own authorship — and it tends to fail on the page too (see below).
4. Ghost-signing / fabrication. Letters the named recommender never saw, or invented recommenders. This is application fraud everywhere, full stop.

The operating principle across every legitimate version: the submitted letter must be the recommender's genuine, informed assessment, submitted by them. Your involvement upstream of that can be debated; anything that breaks that sentence cannot.

How to respond to the ask — four scripts

Redirect to a brag sheet (the default move): "I'm not comfortable drafting the letter itself, but I'll make it really easy — I'll send a one-pager with everything I did under your supervision, dates, outcomes, plus what each program cares about." Most recommenders accept this instantly; it solves their actual problem (recall and time) without transferring authorship.

Clarify the arrangement: "Happy to help — would you want bullet points, or a rough draft you'd rewrite in your own words?" This surfaces whether you are in category 2 or category 3 above. If the answer is "I'll just sign whatever you send," treat it as a yellow flag about how much this letter will help you.

Check the rules out loud: "Let me check the application terms first — a couple of my programs require that recommenders be the sole authors." This is both true diligence and a graceful exit ramp.

Decline and re-source: if the writer's indifference is the message, thank them and quietly find a writer who will actually testify. A lukewarm signer produces a lukewarm letter even when you draft it — you cannot calibrate praise on their behalf, and calibration is what strong letters are made of.

If you do draft it: the craft rules

Suppose you are in the defensible middle of the spectrum — no program bans it, and your recommender has committed to editing and owning the result. Self-drafted letters fail in predictable ways. Avoid these:

Write only what they witnessed. The letter can contain nothing the named writer could not know. If your PI never saw you tutor undergraduates, it does not belong in your PI's letter. Committees notice omniscient recommenders.

Calibrate conservatively. The core of a strong letter is a comparative sentence — "among the strongest N of the X students I have supervised in Y years." When drafting for someone else, you do not know their real numbers. Draft the modest version and flag it: "adjust this comparison to whatever you're actually comfortable signing." An inflated comparison your recommender would not volunteer is exactly the sentence they will get asked about if a program calls to verify.

Build on incidents, not adjectives. The same rule that governs letters written the normal way: one concrete story of what you did, observed from their side of the desk, beats a paragraph of self-praise. Drafting incident-first has a second benefit — it keeps the letter factual, which keeps it honest.

Do not sound like your personal statement. This is the mistake that actually gets caught. Committees read your SOP and your letters side by side, in one sitting. If the letter shares your SOP's vocabulary, pet phrases, and sentence rhythm, the shared authorship is legible on the page. Write the draft in a different register — shorter sentences, evaluator's distance, none of your signature phrases — and never copy-paste between the two documents. (The same logic applies to our statement of purpose vs. personal statement guidance: documents with different jobs should not share a voice.)

Leave room for their edit. Send it as an editable document with a note inviting changes, not a polished PDF. A recommender who changes ten sentences has adopted the letter; one who forwards your PDF has not. You want the former, for your protection and theirs.

Keep it to one or two pages — the norms for every context are here — and let them submit it through their own institutional email, which some universities require outright.

The detection reality

Two separate screening pressures apply to this genre, and both are growing.

First, authorship verification is real: several universities state they may contact recommenders to confirm letters, some use third-party verification, and rescission is the published penalty — the school-by-school enforcement table is here.

Second, stylistic screening is passive but constant. A reader does not need a tool to notice that three "independent" letters and one SOP share a voice — though tools exist too. And self-drafters increasingly compound the risk by drafting with ChatGPT, which layers AI-typical prose on top of voice overlap; universities have said nothing about AI in letters (literally zero of 174 policies address it), but silence is not a safe harbor when the enforcement hook is authorship misrepresentation rather than AI use.

The defense is the same in both cases: a letter grounded in real incidents, calibrated conservatively, voiced unlike your other documents, and genuinely adopted by its signer.

Get the draft reviewed before they sign it

This scenario — you, holding a letter draft you were never supposed to write, with no one you can safely show it to — is precisely what GradPilot's letter review is built for. It reads the draft the way a committee will: relationship evidence, incident specificity, calibration, generic-praise patterns, and AI detection on the full review, so AI-flavored or SOP-flavored prose gets caught before your recommender signs it rather than after a program reads it. Run the draft through GradPilot — free to start. We review the letter; we never write it. That part, for better or worse, is between you and your recommender.

---

Sources

• ProFellow — How to Respond to a Request to Write Your Own Recommendation Letter (79% poll figure)
• Stanford Graduate Admissions — Recommendations (sole-authorship requirement)
• Academia Stack Exchange — "Is it ethical and common that a professor asks the student to write their own rec?"
• U.S. News — Law School Applicants: Don't Write Your Letters of Recommendation
• LSAC — Letters of Recommendation (Credential Assembly Service)