NBME EPC vs Percent Correct: What Your Report Means

Learn how NBME EPC differs from raw percent correct, how passing probability works, and how to interpret Step 1 and Step 2 CK score reports.

Your NBME equated percent correct (EPC) is **not simply the percentage of questions you answered correctly**. Raw percent correct counts correct responses on that specific form; EPC statistically adjusts performance for small differences in form difficulty, while passing probability estimates the chance of passing if you test within one week at a similar knowledge level.

That distinction matters because the numbers answer different questions. Raw percent correct describes what happened on one test form. EPC estimates content mastery on a comparable 0–100 scale. Passing probability addresses near-term readiness—not the percentage of material you know.

EPC, raw percent correct, and passing probability answer different questions

| Report concept | What it means | What it does not mean | |---|---|---| | **Raw percent correct** | Questions answered correctly divided by scored questions on that particular form | A form-adjusted measure that can be compared directly across different NBME forms | | **Equated percent correct** | An estimate of the percentage of the content domain mastered, adjusted for slight differences in form difficulty | The exact percentage of questions you answered correctly | | **Passing probability** | A model-based estimate of passing if you take the corresponding Step exam within one week with a similar level of knowledge | Your EPC, your raw percent correct, or a guarantee that you will pass | | **Likely score range** | An interval showing how your reported result could vary if you retested without meaningful learning or forgetting | A prediction of how much your studying will improve your next score |

NBME explains that equating adjusts scores for slight variations in difficulty so results remain comparable regardless of which form an examinee takes. Because of that adjustment, an EPC may be slightly higher or lower than the literal percentage of items answered correctly. The official CBSSA score-report explanation explicitly distinguishes EPC from unadjusted percent correct.

A raw calculation can still help you review your mistakes. If you answered 150 of 200 questions correctly, your raw percent correct would be 75%. But you should not assume that your official EPC must be 75—or try to reverse-engineer the number correct from an EPC of 75. The relationship depends on the form and the equating process, so there is no universal conversion that learners should apply.

Why NBME uses equated percent correct instead of a raw score

Different forms are designed to assess the same broad content, but they cannot contain identical questions. One form may be slightly more difficult than another even when both follow the same blueprint.

A raw percent correct does not account for that difference. If two learners each answer 70% correctly on forms of slightly different difficulty, those raw percentages do not necessarily reflect identical proficiency. Equating is intended to reduce the effect of form difficulty so that performance can be compared more fairly across forms and over time.

This is why EPC is the better number for following your NBME trend. According to NBME, an EPC represents the percentage of the content domain that would be answered correctly based on the examinee’s proficiency level—not necessarily the exact percentage answered correctly on the administered form.

**Evidence-backed interpretation:** EPC is adjusted for form difficulty and is intended to support comparison across forms.

**Practical recommendation:** Record the official EPC or scaled score shown on each report rather than substituting a manually calculated raw percentage. You can track both, but they serve different purposes.

Step 1 and Step 2 CK reports do not use EPC in exactly the same way

The report structure depends on which comprehensive self-assessment you took.

Step 1 CBSSA reports

For the Comprehensive Basic Science Self-Assessment, the main overall result is a **total EPC score**. The report also includes a likely score range, the Step 1 low-pass range, and an estimated probability of passing Step 1 if you test within one week.

NBME’s guidance for gauging Step 1 readiness says these elements should be considered together. It specifically cautions that overlap between the likely score range and the low-pass range deserves extra caution.

Step 2 CK CCSSA reports

For the Comprehensive Clinical Science Self-Assessment, overall performance is reported on the **three-digit Step 2 CK scale**. EPC is used for the content-area results, allowing you to examine relative performance across areas such as medicine, surgery, pediatrics, and obstetrics and gynecology.

The official CCSSA score-report FAQ explains that the three-digit score estimates how you might have performed on Step 2 CK under the same conditions and at the same knowledge level. The report also includes a likely score range and estimated probability of passing within one week.

This means a Step 2 CK learner should not mistake a content-area EPC for the overall three-digit estimate. They occupy different sections of the report and answer different questions:

• **Overall three-digit score:** estimated Step 2 CK-scale performance.
• **Content-area EPC:** estimated mastery within a content category.
• **Passing probability:** modeled chance of passing within one week under comparable conditions.

A concrete example of reading a Step 1 NBME report

NBME’s sample CBSSA report shows a learner with these results:

• **Total EPC:** 73
• **Likely score range:** 69–77
• **Estimated probability of passing Step 1 within one week:** 99%

Here is the correct interpretation.

The EPC of 73 means NBME estimates that the learner has mastered approximately 73% of the tested content domain after adjusting for form difficulty. It does **not** prove that the learner answered exactly 73% of the questions correctly.

The 69–77 interval represents expected score variation if the learner took another comparable CBSSA without learning or forgetting material. In the sample report, NBME states that a subsequent score would fall within four points of the reported result about two-thirds of the time under those assumptions.

The 99% passing probability is a separate model-based result. It estimates the chance of passing Step 1 if the learner tests within one week with a similar level of knowledge. It does not mean 99% content mastery, nor does it guarantee a pass.

Finally, the low-pass range must be viewed alongside the likely score range. A point estimate can look reassuring while the lower end of its likely range is less comfortable. This is why NBME recommends considering the full report rather than treating one number as a verdict.

Read your NBME report in this order

A consistent sequence prevents you from anchoring on whichever number creates the strongest emotional reaction.

1. Confirm the assessment and testing conditions

Check whether the report is from a CBSSA, CCSSA, CBSE, or CCSE. Then note whether you completed the assessment under standard timing, in one sitting, without outside resources, and without significant exposure to repeated questions.

NBME notes that passing estimates are less reliable when testing conditions differ from the actual Step exam, when resources are consulted, or when a form contains questions previously seen. Treat a self-paced or open-resource result as useful learning data, not as an equivalent test-day simulation.

2. Read the overall result and its scale

For Step 1 CBSSA, begin with total EPC. For Step 2 CK CCSSA, begin with the overall three-digit score.

Do not compare a Step 1 EPC directly with a Step 2 CK three-digit score. Do not compare a content-area EPC with the overall Step 2 CK estimate. First identify the scale, then interpret the number.

3. Locate the likely score range

The range is your reminder that assessment results contain measurement uncertainty. Ask whether the entire range, rather than only the point estimate, sits where you and your school need it to be.

NBME’s Step 2 CK readiness guidance similarly recommends caution when the likely score range overlaps the low-pass range. It also emphasizes that the passing probability should be considered with the other report information.

4. Read passing probability with its time condition

The probability applies to testing **within one week** with a similar level of knowledge. If your exam is several weeks away, it is not a forecast of what will happen after that interval. You may learn more, forget material, become ill, improve your pacing, or test under different conditions.

Also check the model-update date printed on your report. NBME updates its probability models periodically, so the report itself is the correct place to identify which model produced your estimate.

5. Use content EPC to select the next intervention

Content-area EPC is most useful for deciding what to investigate—not for declaring a category permanently “good” or “bad.” Smaller content areas contain fewer questions and generally provide less stable estimates than the overall score.

Before assigning several days to a weak category, inspect the missed questions. Determine whether they reflect:

• a repeated knowledge gap;
• failure to distinguish similar diagnoses;
• incorrect management sequencing;
• misreading or premature closure;
• timing pressure; or
• a small cluster of unrelated misses.

You can then enter your official results into the CoreStepPrep score predictor to organize your trend, while keeping NBME’s own report and probability estimate as the authoritative interpretation of that assessment.

A 48-hour workflow for turning the report into action

This is a report-reading workflow, not a generic study calendar. Its purpose is to move from score interpretation to a small number of defensible changes.

| When | What to do | Checkpoint | |---|---|---| | **First 15 minutes** | Record assessment form, timing mode, overall result, likely range, passing probability, and test date | You can state what each number means without calling EPC “raw percent correct” | | **Next 30–45 minutes** | Compare the likely range with the report’s low-pass range and scan content-area results | You identify uncertainty and avoid relying only on the point estimate | | **First review session** | Classify misses by knowledge, reasoning, management sequence, reading error, or timing | Each priority is tied to actual missed questions | | **Second review session** | Choose no more than two content priorities and one test-taking behavior | Your plan is narrow enough to execute and measure | | **Midpoint before the next NBME** | Test the chosen weaknesses with fresh, timed questions | Accuracy or reasoning improves on unseen questions—not only reviewed items | | **Next comparable assessment** | Compare official scores and ranges under similar conditions | You evaluate a trend rather than demanding a perfect linear increase |

Progress checkpoints that do not invent a readiness cutoff

A universal EPC cutoff would ignore the official report’s likely range, low-pass range, testing conditions, and passing-probability model. Instead of adopting an unofficial threshold, use these checkpoints:

1. **Interpretation checkpoint:** You can distinguish your point estimate, likely range, content EPC, and passing probability.
2. **Validity checkpoint:** Your latest assessment was completed under conditions reasonably similar to test day.
3. **Trend checkpoint:** Multiple comparable reports show stable or improving overall performance rather than one isolated result.
4. **Range checkpoint:** You have examined where the full likely range falls relative to the official low-pass range.
5. **Transfer checkpoint:** Improvements appear on new timed questions, not just during review of familiar material.

A single encouraging score can be meaningful, but it should not erase conflicting evidence such as severe timing problems, repeated forms, open-resource testing, or a previous result obtained only days earlier.

Common score-interpretation mistakes

Treating EPC as an exact question count

EPC may be close to raw percent correct, but it is equated. Do not calculate the number wrong by multiplying EPC by the total number of questions.

Treating passing probability as content mastery

A 90% chance of passing is not 90% mastery. NBME explicitly treats these as separate concepts.

Applying an online conversion to every form

A remembered-items spreadsheet or forum formula cannot reproduce NBME’s equating. Avoid converting EPC into an unofficial three-digit score unless an official report or official table specifically provides that relationship.

Ignoring how the assessment was taken

An assessment completed with pauses, references, repeated items, or substantially extended time may still reveal knowledge gaps. It is simply less representative of test-day performance.

Overreacting to one content-area bar

Use the question-level evidence before rebuilding your plan around one category. A few questions can move a smaller category more than they move the overall result.

Interpreting a probability as a promise

Passing probability summarizes outcomes among examinees with comparable modeled performance under defined timing assumptions. It cannot account perfectly for what happens between the self-assessment and exam day.

Final takeaways

• Raw percent correct is the literal fraction of scored questions answered correctly; EPC is adjusted for small differences in form difficulty.
• Step 1 CBSSA uses total EPC as its primary overall score, while Step 2 CK CCSSA retains an overall three-digit estimate and uses EPC for content areas.
• Passing probability is a separate estimate tied to testing within one week at a similar knowledge level.
• Read the point estimate, likely range, low-pass range, and testing conditions together; do not invent a universal readiness cutoff.
• Use question-level review to decide what to change, then confirm progress with fresh questions and a comparable assessment.

Put the distinction into practice with a fresh NBME-style item: Try a free USMLE question at CoreStepPrep.