Which NBME Form Is Most Predictive of Your USMLE Score?
Scoring
Jul 24, 2026
6 min
You've taken three or four practice forms this dedicated period and they don't agree — one says you're comfortably passing, another says you're borderline. Before you average them together or panic over the lowest one, it helps to know that these forms were never meant to carry equal weight. Some have a published statistic behind them. Most don't. Here's how every active NBME and UWSA form actually ranks, and why.
The Short Answer
For Step 1, UWSA 2 is the closest thing to a validated answer — it's the only form with a published correlation coefficient (R²=0.680) against real Step 1 scores, with NBME 30 as the strongest signal among the NBME-numbered forms. For Step 2 CK, no published study like that exists for any form; community data most consistently points to UWSA 2 and NBME 13 as the strongest predictors, though sources don't fully agree on which of the two is better.
Step 1: Ranking Every Active Form by Predictive Tier
| Tier | Form | Why |
|---|---|---|
| Tier 1 — Validated (published R²) | UWSA 2 | R²=0.680, the highest of any form; overpredicts ~6.9 pts on average |
| Tier 1 — Validated (published R²) | UWSA 1 | R²=0.619; overpredicts ~10.5 pts, a wider margin than UWSA 2 |
| Tier 2 — Strongest community signal | NBME 30 | Hardest-scoring active form; a pass here is the toughest bar to clear |
| Tier 3 — Reliable, known bias | NBME 29 | Tends to underpredict by 5–8 pts — a borderline score may be a real pass |
| Tier 3 — Reliable, known bias | NBME 28 | Tends to overpredict by 3–5 pts — a smaller, more predictable swing |
| Tier 4 — Mid-dedicated checkpoint | NBME 26 | Comparable to NBME 27; no separately tracked bias figure |
| Tier 4 — Mid-dedicated checkpoint | NBME 27 | Comparable to NBME 26; Microbiology-heavy content mix |
| Tier 4 — Mid-dedicated checkpoint | NBME 31 | Newest form; matches current exam style but lacks an established bias pattern yet |
| Tier 5 — Baseline / limited use | NBME 25 | Slightly easier scoring; best for an early baseline read |
| Tier 5 — Baseline / limited use | Free 120 | ±10–15 pt swing, the widest of any form; format rehearsal, not a primary predictor |
Step 2 CK: Ranking Every Active Form by Predictive Tier
| Tier | Form | Why |
|---|---|---|
| Tier 1 — Top community consensus (order disputed) | UWSA 2 | Called the most accurate Step 2 CK predictor in community data; no published study exists to settle it |
| Tier 1 — Top community consensus (order disputed) | NBME 13 | Also called the most predictive CK form; best used 1–3 weeks before the exam |
| Tier 2 — Strong, mid-dedicated | NBME 12 | Explicitly described as having strong predictive correlation with actual scores |
| Tier 2 — Strong, mid-dedicated | NBME 11 | Consistently cited as one of the more reliable mid-dedicated checkpoints |
| Tier 3 — Useful, less discussed as a predictor | NBME 10 | Considered the fairest CK form ever written; community notes praise it for clinical reasoning, not specifically for score prediction |
| Tier 3 — Useful, less discussed as a predictor | NBME 14 | Long HPI-format stems match the real exam; best saved for final prep |
| Tier 3 — Useful, less discussed as a predictor | NBME 15 | Newest CK form; most current content but the least community track record so far |
| Tier 4 — Baseline / limited use | UWSA 1 | Overpredicts by 10–15 pts on average; useful for finding gaps, not scheduling |
| Tier 4 — Baseline / limited use | NBME 9 | Notoriously harsh and outdated; use purely to find gaps, not to judge readiness |
Why the Hardest-Scoring Form Isn't Automatically the Most Predictive
Say you score a 235 on NBME 30 — the hardest-scoring active Step 1 form — but only a 215 on UWSA 2. It's tempting to trust the harder-feeling result more, on the assumption that a tougher form must be a more rigorous measurement. It isn't, statistically. NBME 30 earns its Tier 2 spot because a passing score there is a harder bar to clear, which is a reasonable proxy for readiness — but it has never been checked against real outcomes the way UWSA 2 has. UWSA 2 remains the number to lean on for a scheduling decision, not because it feels more rigorous, but because it's the only one in Tier 2 or below with an actual coefficient behind it. Difficulty is a useful signal. It is not the same thing as validation.
How to Sequence These Forms During Dedicated
Step 1. Run NBME 25 in week 1–2 as your baseline — it's the easiest-scoring form, so treat the number as a floor, not a forecast. NBME 26, NBME 27, and UWSA 1 work as mid-dedicated checkpoints in weeks 3–5; UWSA 1 is the more informative of the three since it's the only one with a published R² behind it. NBME 28 in week 5–6 is a confidence builder — expect it to run a few points generous.
NBME 29 in week 6–7 is your primary readiness check, worth adjusting a few points upward for its known conservative bias, and NBME 30 in week 7 is the strongest NBME-only signal you'll get. In the final stretch, run NBME 31 for the closest match to current exam style, Free 120 for interface rehearsal, and finish with UWSA 2 5–7 days out — the one number in this whole sequence with a peer-reviewed correlation behind it.
Step 2 CK. Start with NBME 9 in week 1–2 purely to find gaps — it's notoriously harsh and outdated, so don't read the score as a forecast. NBME 10 and UWSA 1 in weeks 3–4 test clinical reasoning and set an early benchmark, though UWSA 1's own overprediction (10–15 pts) means it's still for gap-finding, not scheduling.
NBME 11 and NBME 12 in weeks 4–6 are the more reliable mid-dedicated checkpoints, and NBME 13 in week 6 is one of the two top-tier signals for CK. In the final stretch, NBME 14 and NBME 15 match current exam style and content, and UWSA 2 7–10 days out is the other top-tier signal — run it last so it's the number closest to your test date.
Using the Ranking, Not Just Reading It
A ranking is a starting point, not a substitute for your own trend. If UWSA 2 and NBME 30 agree in the low 210s for Step 1 — or UWSA 2 and NBME 13 agree in the low 230s for Step 2 CK — that agreement across tiers is a stronger signal than either score sitting alone. For the full explanation of what R² actually means and how to adjust each form's known bias before trusting a number, see NBME vs USMLE Score Correlation.
Bottom line: treat Tier 1 forms as your scheduling anchor, everything else as supporting evidence, and don't let a single hard-feeling score outrank a validated one.
Frequently Asked Questions
References
Community Data Sources
This article is for educational purposes only. Not affiliated with NBME® or USMLE®. Predictions and score estimates carry an estimated error of ±5–10 points and do not guarantee a passing result, reported score, exam outcome, or eligibility decision. Where cited, correlation data comes from peer-reviewed research (see References above); everything else reflects community-reported patterns (see Community Data Sources above), not a formally published statistic.