A commonly cited rule of thumb is a UWorld percentage in the high 60s or better as dedicated study progresses, and above the mid 70s is strong. The problem is that this number, on its own, is nearly meaningless, because how you generate it changes it by 15 points or more. Your cumulative percentage is a useful learning signal and a poor score predictor, and confusing the two is the single most common benchmarking mistake. Here is what the number actually means and what to track instead.
Key takeaways
- A UWorld percentage in the high 60s and up is a common benchmark, but context changes it enormously.
- Percent correct and percentile are different numbers that students constantly conflate.
- Five factors distort your percentage: reset, mode, question age, block mix, and review habits.
- NBME self-assessments and the Free 120 predict your real score best; raw percentage predicts worst.
- The real exam has measurement error of about 6 points, so obsessing over one mock is misguided.
The direct answer, and why it is almost useless alone
As a rough benchmark, many students aim for a first-pass UWorld percentage in the high 60s or higher during dedicated study, trending upward toward the exam. But a raw percentage is contaminated by how you produced it. Two students both sitting at 68 percent can be 15 points apart on the real exam, because one built that number on a fresh first pass under timed, random conditions and the other on a second pass of previously seen questions in tutor mode. The number is not wrong, it is just uninformative until you know how it was made.
Percent correct versus percentile
These are two different numbers, and people mix them up constantly. Your percent correct is how many questions you answered correctly. Your percentile is how you rank against other users, which UWorld also reports. A 70 percent correct is not a 70th percentile, and neither directly equals a scaled Step 2 CK score. When someone asks "is 65 percent good," the honest response is that it depends what you mean and how you got it, because the two numbers answer different questions.
Five things that distort your percentage
Before you read anything into your percentage, account for these:
- First pass versus reset. A second pass over questions you have seen inflates your percentage well above your true first-encounter performance.
- Tutor mode versus timed mode. Tutor mode, with immediate feedback and no clock, produces higher percentages than timed, exam-like blocks.
- Question age in your subscription. Older questions you half remember lift your number without reflecting new learning.
- Block composition. Organ-system or subject blocks you just studied score higher than random, mixed blocks that mirror the real exam.
- Review behavior. Whether you actually learn from explanations or just click through changes what the percentage means going forward.
Read your percentage only after correcting for these, and always weight fresh, timed, random blocks most.
What actually predicts your score
There is a clear evidence hierarchy for Step 2 CK prediction. At the top sit the official NBME self-assessments and the Free 120, because they use the same item family and the same scale as the real exam, so they translate most directly. Next come the UWorld self-assessments, which correlate well but carry their own biases. At the bottom sits your raw question-bank percentage, which correlates only loosely with the real score. The practical rule follows: benchmark with NBME forms, use UWSAs as a secondary check, and treat your cumulative percentage as a study signal rather than a forecast.
Measurement error, and why single-mock obsession is misguided
The real exam is not a point estimate. Step 2 CK has a standard error of measurement of about 6 points, which means your true ability sits within a band, not on a single number, and a practice test predicts a range rather than an exact score. Agonizing over a 3-point difference between two mocks, or between a mock and your goal, is reading noise as signal. Think in bands: a predicted 248 realistically means somewhere in the low-to-high 240s, and your preparation should aim to move the whole band, not chase a specific digit.
A sane benchmarking protocol
Keep it disciplined. Take a baseline NBME early so you have a real starting point, then trend your scores over time rather than fixating on any single result. Sit predictive assessments under fixed, timed, exam-like conditions so they are comparable, and take your most predictive forms in the final one to four weeks, when they best reflect where you are. Once two or three recent scores cluster, stop adding mocks, because more numbers past that point add anxiety, not information. Trajectory beats any one data point.
An honest note on resources
UWorld is an excellent core question bank, and its depth of explanation is why it is the standard. Its percentage is a good study signal; it is just not a score predictor, and that is a property of raw percentages, not a flaw in UWorld. Adaptive engines solve a different problem: rather than being a primary content bank, they target the specific concepts you keep missing and schedule retention, which complements a core bank rather than replacing it. iatroX is built for that targeting-and-retention job, and you can judge the fit with its free sample questions at iatroX. For the fuller prediction picture, see can a question bank predict your Step 2 CK score, and for the numbers that matter this year, the 218 standard and the 250 mean.
Frequently asked questions
What is a good UWorld percentage for Step 2 CK? As a rough benchmark, the high 60s or better on a first pass during dedicated study, trending up toward the exam. But the number depends heavily on mode, pass, and block composition, so read it in context.
Does my UWorld percentage predict my Step 2 CK score? Only weakly. Raw percentage correlates loosely with the real score. NBME self-assessments and the Free 120 predict best, with UWorld self-assessments next. Use your percentage as a study signal.
Is percent correct the same as percentile? No. Percent correct is how many you got right; percentile is how you rank against other users. Neither equals a scaled Step 2 CK score, and they are frequently confused.
How accurate are practice tests? They predict a range, not a point. Step 2 CK has a standard error of measurement of about 6 points, so treat a predicted score as a band and avoid over-reading small differences between mocks.
How should I benchmark my prep? Baseline early with an NBME, trend over time under fixed timed conditions, take your most predictive forms in the final one to four weeks, and stop testing once recent scores cluster.