Table of Contents
A few days ago, a student sent this message:
“Got 63% on NBME 31. Took Free 120 two days later and got 76%. Did I actually improve that much, or is one of these just easier? My exam is next week.”
It wasn’t the first message like this. In the final 10 days of dedicated prep, this is one of the most common panic patterns — the sinking feeling after NBME 31, some confidence returning after Free 120, then a new and somehow worse confusion: which number do I actually trust?
The honest answer is both — but for different things. This article explains exactly what each exam is measuring, why a 10–15 point gap between them is mathematically normal, and what to actually do with both scores in your final week.
The context most score-comparison articles skip entirely
On January 26, 2022, Step 1 transitioned to pass/fail. The intention made sense — reduce the unhealthy dominance of a single numeric score in residency selection, encourage a more holistic evaluation of applicants.
The data that followed was uncomfortable.
88% → 82%
Overall Step 1 pass rate dropped in a single year
US MD students: 96% → 93% · US DO: 94% → 89% · IMGs: 82% → 74%. The decline continued into 2023. Researchers noted that reduced preparation intensity — a “just need to pass” mindset — was a likely contributor.
Alongside the pass/fail change came another detail that rarely gets mentioned: the passing threshold increased from 194 to 196. That’s a small number, but it sits right in the middle of the score distribution where thousands of students cluster.
The result is that score anxiety didn’t disappear — it relocated. The old pressure was “will I hit 240?” The new pressure is “will I pass at all?” And in many ways, the second question carries more weight. That’s why the Free 120 vs NBME 31 comparison matters: misreading either number has real consequences.
NBME 31: it’s not harder — it’s newer
The most common misconception about NBME 31 is that it’s deliberately brutal. It isn’t. It’s the most recently released CBSSA form, and that single fact explains almost everything that feels different about it.
Newer forms — 28, 29, 30, 31 — are more closely aligned with the current exam blueprint than older forms (15, 16, 17), which now tend to overpredict actual scores by 5–15 points because the content distribution has shifted since their release. Forms 30 and 31 show the tightest correlation with actual Step 1 performance of any available practice forms.
The long stems, the vague mechanism questions, the “all of these are technically correct” answer choices — that’s what the real exam currently looks like. The unfamiliarity isn’t a flaw in the form. It’s the point.
Score estimate
The community-derived formula used by most tutors and calculators:
Getting 130–135 questions correct (65–67.5%) puts most students in the 208–212 range — safely above the passing threshold with margin for test-day variance.
A note on burnout and misread stems
There’s an underreported phenomenon worth knowing: students who have been studying 10–12 hours a day for weeks often report “bombing” Form 31, then discover in review that they knew the answers — they misread stems or lost focus on a single key word. Cognitive fatigue is real. A 24-hour proper rest before review changes the picture significantly more often than people expect.
Free 120: what NBME actually designed it for
NBME did not design the Free 120 as a score-prediction tool.
It provides a raw percentage — no three-digit scaled score, no likelihood-to-pass estimate. NBME’s explicit stated purpose is exam interface familiarization and format practice. Blueprint Prep’s medical education team puts it plainly: it “shouldn’t be used as the sole indicator of readiness for the USMLE.”
The reason students use it as a score proxy is understandable — there’s no other free, official practice available in the final few days, and anxiety peaks at exactly that moment. But being honest about what the percentage actually represents is essential to interpreting it correctly.
Community benchmarks — with honest caveats
The thresholds most commonly cited by Step 1 educators:
- 70%+ — comfortable territory, passing highly likely
- 65–69% — generally safe, consistent with NBME passing-range performance
- Below 60% — worth taking seriously; cross-reference with recent NBME trend
These numbers come from educator anecdotal data based on approximately 25–30 students. No large-scale peer-reviewed study has validated specific Free 120 percentage thresholds. Treat them as directional, not definitive.
On the question of Free 120’s statistical correlation with actual Step 1 scores: no peer-reviewed study has published a verified r-value for the current version of the Free 120. Any article citing “r ≈ 0.65” or similar without a source is presenting an unverified claim. The honest answer is that the correlation is moderate — enough to be informative, not tight enough to be reliable on its own.
The UWorld familiarity effect
Published PMC research analyzing 399 students (Rao et al., 2020) found that UWorld practice test scores were the most predictive of actual Step 1 performance — outperforming NBME forms in predictive accuracy. Free 120 questions are written in a UWorld-adjacent style. If you’ve been UWorld-heavy throughout dedicated, Free 120 will feel familiar — not because your knowledge exploded in 48 hours, but because the question format matches what you’ve been doing for weeks. That familiarity explains a significant portion of the gap between your NBME 31 and Free 120 scores.
Visual Reference — Free 120 vs NBME 31
NBME 31
CBSSA — current blueprint
Free 120
Interface practice tool
NBME Forms 29, 30, 31 — Score Zone Reference
Community consensus, not NBME-official · Safe zone = 210+ on two consecutive recent forms · ~60–65 wrong out of 200
Passing score (since 2022)
196
Was 194 before Jan 2022
Free 120 safe threshold
65%+
Anecdotal · ~25–30 student sample
Official score variance (SEE)
±7–8
USMLE official per attempt
Why a 10–15 point gap is statistically normal
The USMLE’s official Standard Error of Estimate (SEE) for Step 1 is approximately ±7–8 points. This means that if the same student took the same exam three times, two of those three scores would fall within a ±7–8 point range. That’s not your knowledge fluctuating randomly — it’s the inherent statistical variance built into any standardized test.
Now consider that Free 120 and NBME 31 differ in every variable that affects test performance: total question count (120 vs 200), scoring output (percentage vs scaled estimate), question style, stated purpose, and where they typically fall in the study timeline. A 10–15 point gap isn’t evidence that something went wrong. It’s the expected result of two different instruments measuring two overlapping but distinct things.
The gap becomes genuinely concerning only when both scores are consistently below passing territory, and your recent NBME trend confirms it. One bad day doesn’t define your trajectory.
One practical note: if you didn’t take Free 120 under strict timed conditions — extra breaks, phone checks, pausing — your percentage is inflated. Strict conditions are non-negotiable for any meaningful comparison.
Final week decision framework
Here’s exactly what to take, when, and why — no ambiguity.
7–10 Days Before Exam
Take NBME 31 under strict timed conditions — 1 hour 15 minutes per section, no extra breaks. Use the formula or a community calculator for your score estimate. This is your primary content calibration tool.
3–5 Days Before Exam
Take Free 120. Primary goal: interface practice and final confidence calibration. Score is secondary. Do not directly compare the percentage with your NBME 31 result — they’re different instruments.
Borderline NBME 31 Score (208–214)
One form’s data is insufficient to make a decision. If 2–3 recent NBME forms are consistently in passing territory, trust the trend over any single snapshot. Trend data is more predictive than a single form.
Never
Don’t take both exams on the same day. Seven full blocks in a single session produces severe cognitive fatigue that directly hurts performance — not just morale.
Safe zone benchmarks
| Exam | Threshold | What It Signals | Caveat |
|---|---|---|---|
| NBME 29/30/31 | 210+ on 2 consecutive forms | Safe zone — community consensus | ~68–70% correct, ~60–65 wrong out of 200 |
| Free 120 | 70%+ | Comfortable territory | Anecdotal data, ~25–30 student sample |
| Free 120 | 65–69% | Generally safe | Cross-check with recent NBME trend |
| Any recent form | Below 60% | Red flag — strategy review | Especially if consistent across 2+ forms |
Published PMC research analyzing 399 students (Rao et al., 2020, PMC7198101) found that taking six or more practice tests significantly enhanced Step 1 scores, with an optimal range of 6–9 total tests and diminishing returns beyond that. Don’t stack everything into the final week — spread practice tests across your dedicated period.
On trend: a student who scores 210 → 225 → 238 across three sequential forms is on a trajectory that predicts closer to 245–250 on the actual exam — not 238. Trajectory outperforms any single snapshot.
Frequently asked questions
Que. Do Free 120 questions actually repeat on the real exam?
Ans. Occasionally, yes. Students report seeing 1–2 nearly identical questions. That makes reviewing your Free 120 incorrects genuinely valuable in the final days — it’s not just confidence practice, it’s potential direct ROI on test day.
Que. Why does NBME 31 feel so unfamiliar compared to older forms?
Ans. It’s the newest form, which means it has the most recently updated question writing style. That unfamiliarity creates an illusion of difficulty — the format feels harder because it’s newer, not necessarily because the content is harder. If you’re hitting 130–135 correct, you’re in the safe passing zone regardless of how it felt.
Que. I got 62% on NBME 31 and 75% on Free 120. What should I do?
Ans. NBME forms are the backbone of any score prediction. Free 120 is a sanity check. If your last two NBME forms both predict passing territory, a 75% Free 120 confirms it. What matters most: are multiple recent NBME forms pointing in the same direction? If yes, proceed.
Que. How many practice tests should I take in total?
Ans. Published research (Rao et al., 2020, PMC7198101) found the optimal range is 6–9 total practice tests, with significant score improvement up to that point and diminishing returns beyond it. Spread them across your dedicated period rather than clustering everything in the final week.
Que. Should I trust NBME 31 or Free 120 more for pass/fail prediction?
Ans. NBME 31 — it’s designed for score prediction and gives you a scaled estimate with a likelihood-to-pass percentage. Free 120 gives you raw percentage and is explicitly designed for interface practice, not prediction. Use NBME 31 as your primary data point and Free 120 as supplementary confirmation.
What the numbers are actually telling you
Neither exam is giving you a wrong number. They’re just measuring different things — and the gap between them trips people up because it looks like a contradiction when it isn’t one.
NBME 31 tells you where you stand relative to the current exam’s difficulty. Free 120 tells you how you execute when everything looks and feels like test day. You need both data points. But NBME is your foundation — Free 120 confirms it, it doesn’t replace it.
The student who messaged me — 63% on NBME 31, 76% on Free 120 — had completed six weeks of dedicated prep, taken three NBME forms, and finished UWorld once. His last two NBME forms were both in passing territory. I told him to proceed. He did. He passed.
One number doesn’t define your outcome. Look at your trend across multiple forms, take both exams under strict conditions, and if the data is consistently pointing toward passing — trust that data, not a single off day.
Sources
- English K et al. “Assessing the Impact of USMLE Step 1 Going Pass-Fail: A Brief Review of the Performance Data.” Avicenna Journal of Medicine, PMC11896725, December 2024.
- Rao S et al. “Correlation of MCAT Scores and Self-assessment Materials with USMLE Step 1 Performance.” PMC7198101, 2020. — ncbi.nlm.nih.gov
- USMLE Official 2022 Step 1 Performance Data. — usmle.org
- Blueprint Prep / Dr. Mike Ren. Free 120 educator analysis. — blueprintprep.com
- NBME 31 score formula: community-derived regression estimate — not officially published by NBME.
NBME® and USMLE® are registered trademarks of their respective organizations. This article is not affiliated with or endorsed by NBME or USMLE.
