USMLE Score Predictor for Step 1, Step 2 CK and Step 3

Each row of the USMLE score predictor is one assessment: the provider, the score it reported, and how many days ago you sat it.

Six providers feed the USMLE score predictor: NBME self-assessments, UWorld SA 1 and SA 2, the Free 120, AMBOSS self-assessments and CMS forms.

Percent-scored forms are mapped onto the three-digit scale first, so 78% on the Step 1 Free 120 enters the USMLE score predictor at roughly 206, and 75% on the Step 2 CK Free 120 at roughly 232.

Results that already report three digits reach the USMLE score predictor exactly as typed. Nothing is rounded away, rescaled, or quietly dropped.

Four things. A point estimate, a ± interval, the pass probability for that exam’s standard, and each row’s share of the USMLE score predictor blend.

That share column is where the USMLE score predictor shows its working. One recent form often carries half the weighted prediction on its own.

Read the interval before the headline number. A USMLE score predictor that hands you a single figure with no width has hidden the part that decides whether to sit the exam.

This is a weighted mean, not a flat average. The USMLE score predictor scales every result by 1 ÷ (1 + days ÷ 10) before blending it.

Sat today, a form carries a weight of 1.00 in the USMLE score predictor. Ten days back it carries 0.50, a month back 0.25, two months back 0.14.

The difference is not cosmetic. An NBME of 218 sat thirty days ago beside a UWorld SA 2 of 236 sat five days ago averages 227, but the USMLE score predictor returns 231.

Swap the dates and those same two numbers come back 223, because the USMLE score predictor is reading your score trend rather than your arithmetic mean.

Nothing is ever deleted. Old sittings keep a voice in the USMLE score predictor; recency weighting just stops them dominating a run of newer work.

Step 1 has been reported pass or fail since January 2022, so a USMLE score predictor number is not something you can put on an application.

What it is good for is distance from 196, the Step 1 passing standard, which this USMLE score predictor turns into a pass probability.

That curve moves fast near the line. In the USMLE score predictor a single NBME at 196 returns exactly 50%; 200 returns 69.9%, 205 returns 88.0%, 210 returns 96.6%.

Fourteen points separate those four readings, and the USMLE score predictor still prints a ±15 confidence range beside every one of them.

More inputs tighten it. Three Step 1 forms — 208 forty days ago, 214 twenty days ago, a UWorld SA 1 of 219 three days back — leave the USMLE score predictor at 216, ±12, so 204 to 228.

Step 2 CK still reports a three-digit score, so the USMLE score predictor point estimate matters as much as the margin. The passing standard here is 218.

A single NBME entered at 218 leaves the USMLE score predictor at exactly 50%, which is what sitting on the standard actually means.

One assessment at 225 takes the USMLE score predictor to 82% and a range of 210 to 240: clear of the line on the estimate, still touching it at the bottom.

That overlap is the argument for sitting another form rather than booking the date.

Step 3 runs through the USMLE score predictor against a 200 standard, in force since 1 January 2024, with identical weighting and interval arithmetic.

It is the target most people hold the fewest inputs for, so the USMLE score predictor usually rests on one or two rows here and stays near the ±15 floor.

An offline USMLE score predictor cannot audit itself against real exam outcomes, so the width of the interval is the accuracy claim.

One assessment produces a ±15 point range, and that is the USMLE score predictor floor. No amount of confidence in a single sitting narrows it.

Agreement is what the USMLE score predictor rewards. Two results at the same score on the same day return ±11; three return ±9.

Disagreement is what it punishes. Two scores ten points apart return ±14, and a 205 beside a 255 returns ±50 from the USMLE score predictor.

So when your results split by more than ten points, the useful question is why, not what the average of them is.

You will not find a validation statistic here. Other sites quote correlations from cohorts of their own; those are their figures, and this USMLE score predictor does not borrow evidence it has not collected.

The conversions under this USMLE score predictor come from publicly circulated data built on community-reported results, not from the exam makers.

NBME publishes no formula turning a self-assessment into a three-digit score, so every offline USMLE score predictor, this one included, is a reconstruction rather than an official reading.

The USMLE score predictor also knows nothing about your exam day. Content sampling varies between forms, fatigue lands at hour six, and a bad night before the test shows up in none of the numbers you typed.

It cannot tell whether you sat a form timed or with the clock paused, or whether a high score means mastery or a form you had already seen.

Use the USMLE score predictor for direction and readiness rather than for one high-stakes decision. Two forecasts six weeks apart say more than either one alone.

Holding one result only? A converter answers that better than any USMLE score predictor can — start from a single form instead.

Or go straight to Step 1 forms 25 to 31 and Step 2 CK forms 9 to 15, then bring the numbers back here.

Once the forecast is one you believe, turn the prediction into a schedule, or read the weighting rules in full.