Step 3 self-assessment tool
Enter your number incorrect or percent correct to estimate an NBME 6 scaled score. The conversion is a transparent community-data approximation, not an official NBME report or a USMLE Step 3 prediction.
NBME 6 Step 3 calculator
Choose wrong answers or percent correct. The tool converts the input to a 200-question missed count and applies a simple regression fitted to publicly discussed NBME 6 results.
Estimated NBME 6 scaled score
--
Independent estimate
NBME does not publish the community regression used here. The calculator estimates an NBME 6 self-assessment scaled score from question performance; it does not reproduce the official scoring process, predict an exact Step 3 score, or replace guidance from NBME, USMLE, your residency program, or a physician educator.
Method and assumptions
Searchers often want a fast conversion after completing Form 6 offline. This page completes that task while showing the assumptions, boundaries, and practical uses that affect the result.
Use the number incorrect shown in your review notes, or enter percent correct. Percent mode first converts the percentage to a missed-question estimate out of 200. Avoid mixing the official NBME scaled score with a raw percentage; they are different inputs. If an offline file or review source uses another denominator, calculate the percentage first rather than entering an unadjusted count.
The estimate uses a linear regression fitted to four publicly discussed Form 6 result pairs: 110 correct / 267, 120 / 337, 131 / 415, and 143 / 508. In that narrow observed range, each additional correct answer changes the fitted center by about 7.29 scaled points. The sample is small and cannot reproduce official equating, unscored items, or reporting precision.
The calculator displays a ±30-point interval because four community-reported pairs cannot model official equating. Rounding, recall errors, offline-file differences, and testing conditions can all move the estimate.
The output is an estimated NBME 6 self-assessment scaled score. It is not the USMLE Step 3 three-digit score. Community discussions sometimes compare the two, but there is no official one-number conversion that works equally for every examinee. Do not subtract, divide, or match the output to a passing standard as if the scales were identical.
A safer Step 3 decision includes the official NBME report, question-block performance, CCS practice, timing, fatigue, clinical knowledge weaknesses, and program guidance. One converted form should not be the only reason to keep, move, or cancel an exam date. Look for consistent evidence across several recent and realistically timed activities.
The calculation runs in your browser. You do not need to enter a name, NBME account, registration number, residency program, diagnosis, or other medical information. Record only the numerical result you need for your study plan. If you share a screenshot, remove any identifiers from the surrounding score report.
A simple regression is easier to explain and audit than a black-box prediction. The page states the assumed question count, shows an uncertainty range, and avoids claiming that hidden official data were used. That transparency is more useful than displaying extra decimal places that the source data cannot support.
Do not force NBME Forms 5, 7, or 8 into this regression. Each form may have a different difficulty distribution and reporting relationship. Use the separate Form 7 calculator for Form 7 results, and use an official report whenever one is available.
Result guide
The bands below are editorial study-planning labels. They are not official pass probabilities, score categories, licensing advice, or a replacement for the interpretation included with your official self-assessment.
Use this as a prompt to inspect broad weaknesses rather than as a verdict. Review the official content-area feedback, identify repeated misses, and distinguish knowledge gaps from rushed reading. A second checkpoint after targeted work is more informative than repeatedly recalculating the same form.
A developing result can contain both solid and unstable domains. Separate clinical knowledge gaps from pacing, fatigue, or question-reading errors. Compare the estimate with CCS performance and a later assessment before making a high-stakes scheduling decision.
This range is near the conventional center of the NBME-style scale, but it still does not guarantee a Step 3 outcome. Stability across question blocks, CCS cases, and recent practice matters more than the label alone. Continue reviewing misses rather than treating the number as permission to stop.
A stronger estimated result is encouraging, yet the regression can overstate precision at the extremes where public data are sparse. Confirm the result with the official NBME report and preserve the study habits that produced it. Do not use a high estimate to ignore major CCS or timing weaknesses.
Percent-correct mode rounds to an estimated missed count. If you know the exact number wrong, use it. One question moves this fitted center by about 7.29 points, but that mathematical change should not be treated as official measurement precision.
Do not use it for a different NBME form, an official score already shown on your report, UWSA, Free 137, CCS cases, an in-training examination, or the real USMLE Step 3 exam. Those products use different structures and scales. Use the official result whenever it is already available.
Practical examples
These examples show how to choose the correct input and how much confidence to place in the output.
| Situation | Input | Best use |
|---|---|---|
| You counted 58 missed questions | 58 wrong | Use the exact missed count for the least rounding and keep the estimated range visible. |
| You recorded 71% correct | 71 percent | The tool estimates about 58 wrong out of 200 before applying the regression. |
| Your official report shows a scaled score | Official score | Use the official score. Re-entering it as a percentage or wrong count would be invalid. |
| You want an exact Step 3 prediction | Converted estimate | Do not treat NBME 6 as a one-to-one USMLE prediction or pass guarantee. |
| You are comparing Forms 5, 6, 7, or 8 | Separate reports | Compare trends cautiously because forms and reporting contexts differ. |
| You completed the assessment untimed | Raw result | Interpret it more cautiously because timing changes how well the session resembles test conditions. |
Official context
Official pages explain the assessment and USMLE reporting context. They do not endorse this community formula.
Related tools
Use these pages for Step 1, Step 2 CK, Free 120, or a broader study schedule. Their score scales and formulas are separate from NBME 6 Step 3.
NBME 6 FAQ
No. It is an independent educational estimate based on a small set of publicly discussed results. NBME and USMLE do not publish or endorse this regression.
It outputs an estimated NBME 6 self-assessment scaled score and a broad uncertainty range. It does not output an official USMLE Step 3 score.
It assumes 200 questions for the raw-performance conversion. If your source or report uses a different denominator, do not enter the count without adjusting it.
Yes. Percent correct is converted to an estimated missed count out of 200, so small rounding differences are expected.
The regression is based on only four publicly discussed Form 6 result pairs. The wider range communicates model and reporting uncertainty instead of presenting a false exact score.
No. NBME self-assessment scaled scores and USMLE Step 3 scores are different scales. Never read the output as a one-to-one Step 3 score.
Not reliably by itself. Use the official NBME report, CCS readiness, recent question performance, and program guidance.
Use the official score. This calculator is mainly for examinees who have only a raw wrong-answer count or percent correct.
No. Form 7 has a separate conversion page and a different community-data regression. Use the calculator that matches the form you completed.
The calculator runs in the browser and does not require identifying information. Avoid entering names, IDs, or private medical details.