Med School Rec Letter Phrases: Strong vs Weak

Med School Rec Letter Phrases: Strong vs Weak

No phrase makes a medical school letter strong on its own. A strong phrase carries one observed incident and one named comparison group. "I recommend her without reservation" with neither is weaker than "one of the two strongest of the 35 research assistants I have supervised" with both.

Lists of "power words" miss this. A reader cannot check your adjectives. A reader can check whether the sentence next to the adjective shows the applicant doing something, and whether your ranking says who you ranked them against. This page works at the level of the sentence: openings, tier words, faint praise and gendered wording, each with a weak and a strong version. For the letter's overall shape, start with how to write a letter of recommendation for medical school.

The TMDSAS evaluator guidance and the two published studies quoted below were checked on November 7, 2026. The AAMC's letter-writer guidelines page returned HTTP 403 to our checks that day, so this page does not quote it.

What letter readers say they want

The clearest published instruction to medical school letter writers that we could load today comes from TMDSAS, the Texas application service. Its evaluator support page says:

"The admission committees are looking for candid, honest insight into the applicant's strengths and weaknesses. A substantial evaluation provides an accurate assessment of the applicant's suitability for dental, medical, and podiatry school rather than advocate for the applicant."

It then asks for "your description of how the applicant has, or has not, demonstrated" a list of competencies, including "Reliability and Dependability," "Teamwork" and "Capacity for Improvement." On the TMDSAS evaluation form, the metrics box says: "please refrain from relisting grade performance."

One study tested which letter features predicted later performance. DeZee et al. (Academic Medicine, 2014) coded 437 admission letters at one school, the Uniformed Services University, against 76 characteristics. Three survived their regression model:

"Being rated as ‘the best’ among peers and having an employer or supervisor as the LOR author were associated with induction into AOA, whereas having nonpositive comments was associated with bottom of the class students."

The scope is one institution and three graduating classes (2007 to 2009), and the authors conclude that "very few LOR characteristics predict how students perform during medical school." Two of the three surviving features are about wording: a ranking against peers, and nonpositive comments. That is the basis for the rest of this page.

Our view: write every evaluative phrase as a claim the reader could test: what you saw, and compared with whom. The countercase: a letter writer with a truly brief vantage, such as one large lecture course, cannot supply incidents, and stacking strong phrases on a thin basis makes the gap more visible. That writer should say what they saw and stop, or decline the request.

Openings: the first sentence sets your vantage

The opening tells the reader how much weight to give everything after it. "It is my pleasure to recommend" tells them nothing. Two constructed examples, written by us, not real recommendation letters:

The strong versions give duration, capacity and how directly you watched. They also make every later claim checkable against that vantage. The "120 shifts" is part of the constructed example, not a statistic.

The tier words: "highly," "without reservation," "strongest"

Writers ask whether "I recommend without reservation" outranks "I highly recommend." None of the admissions pages quoted here publishes a phrase scale, so no published source ranks the two. What you control is whether the phrase has support.

Constructed tier table, written by us:

Our view: use the highest tier your evidence carries and no higher. If you write "the best," name the group and its size. DeZee's one-school finding is about a "the best" rating, and a rating with no group behind it gives a reader nothing to weigh.

The countercase: some letters must fit a fixed form. The TMDSAS evaluation form gives "12 open response text boxes" and says "all text boxes have a 2000 character limit." Inside a 2,000-character box, a short tier phrase plus one incident is the whole answer. Do not spend the box on adjectives.

Faint praise and red-flag phrases

The DeZee study's third surviving feature was "nonpositive comments." The examples below are our own reading of how each phrase lands, not findings from that study. Each one is a hedge a writer may not intend.

A real weakness is different from a hedge. TMDSAS asks for "strengths and weaknesses." If you name one, name it once, show what the applicant did about it, and keep it out of your closing line. The guide to weakness, hardship and disability in a med school letter covers that sentence, and when the applicant's consent comes first.

Gendered and coded wording

The published evidence here covers residency and faculty-hiring letters in medicine, not medical school admission letters. Khan et al. (Postgraduate Medical Journal, 2023) reviewed 16 studies covering 12,738 letters for 7,074 applicants. They report that "64% (7/11) studies found a significant difference in gendered adjectives between men and women", and that in 6 of those 7, "women applicants were more likely to be described using communal adjectives, such as ‘delightful’ or ‘compassionate’, while men applicants were more likely to be described using agentic adjectives, such as ‘leader’ or ‘exceptional’." They also noted "more frequent use of doubt raisers and mentions of applicant personal life and/or physical appearance" in letters for women.

Two constructed examples, written by us, not real recommendation letters:

Weak: "Hannah is a warm, delightful presence on the unit, and patients love her."

Strong: "Hannah ran the unit's falls-prevention huddle for six months. She chose the three patients to discuss each morning, and the charge nurse adopted her checklist for the night shift."

The strong version is not "agentic" by swapping adjectives. It is a decision the applicant made and a result someone else acted on.

The swap test: before you upload, change the applicant's name and pronouns to another gender and read the letter again. Every adjective that now sounds odd, and every sentence about appearance, family or personality that would not appear in the other version, is a sentence to replace with an incident.

A phrase audit before you upload

1. The first sentence states how long you have known the applicant, in what role, and how directly you observed them.
2. Every adjective has an incident within two sentences, with the applicant's own action visible.
3. Every ranking names the group, its size and your basis. If you cannot, delete the ranking.
4. Search the draft for "solid," "mostly," "generally," "seems," "more … than most," "did what was asked" and "no reason to doubt." Each one is either a deliberate, explained weakness or it goes.
5. No grade is restated without context the transcript lacks.
6. The swap test changes nothing.
7. The closing endorsement is the highest tier the body supports, and it is the last sentence of the letter.

Questions about medical school letter phrases

Is "without reservation" stronger than "highly recommend"?

No published scale says so. Each is as strong as the evidence beside it. "Without reservation" also tells the reader to look for reservations, so a hedge anywhere in the letter undercuts it.

What is a red flag in a medical school letter of recommendation?

Phrases that describe compliance ("did what was asked"), rank without a group ("one of the better students"), or doubt dressed as praise ("no reason to doubt"). The DeZee study found that nonpositive comments were associated with students who finished at the bottom of their class at one school.

How should a medical school letter open?

With your vantage: how long, in what role, and how directly you watched. A strong sample opening is in the table above, and a full constructed letter is in the physician letter of recommendation sample.

Should I use the competency names in the letter?

You can, but the incident does the work. TMDSAS asks how the applicant "has, or has not, demonstrated" each competency, so "Reliability and Dependability" with no shift, deadline or task behind it adds nothing. The premed competencies page for letter writers maps incidents to each one.

Get a free review of the letter

Get a free Quick Review of the letter against GradPilot's medical school letter of evaluation rubric. It checks whether each quality you claim has an incident behind it, whether your comparisons carry their own yardstick, and whether stock or hedged phrasing undercuts your closing endorsement. The same checks for a PA, dental or pharmacy file run on the health professions evaluation letter rubric.

More guides for medical school letter writers:

• How to write a letter of recommendation for medical school: the four parts of the letter and what to leave out.
• Medical school letter of recommendation sample from a physician: a full constructed letter, annotated.
• Weakness, hardship and disability in a med school letter: what to say with the applicant's consent.
• All medical school essay and letter guides.

Sources

Retrieved November 7, 2026, except where noted.

• TMDSAS, Evaluator Support (no cycle year stated; letter date preference "after May 1, 2025"): "candid, honest insight," "rather than advocate," the competency list, the 12-box form with a 2000-character limit, "refrain from relisting grade performance."
• DeZee KJ et al., What aspects of letters of recommendation predict performance in medical school? Findings from one institution, Academic Medicine 2014;89(10):1408-15 (classes of 2007 to 2009, one institution): 437 letters, 76 characteristics, three predictive features. Abstract read through the NCBI PubMed record.
• Khan S et al., Gender bias in reference letters for residency and academic medicine: a systematic review, Postgraduate Medical Journal 2023;99(1170):272-278 (studies to July 2020; residency and faculty-hiring letters): 16 studies, 12,738 letters, gendered adjectives and doubt raisers. Abstract read through the NCBI PubMed record.
• Not retrieved: AAMC, Guidelines for Writing a Letter of Evaluation, HTTP 403 on November 7, 2026. Not quoted.