How NBME Shelf Scores Work

How do you interpret a shelf exam score? The NBME shelf exams were originally scaled to a mean of 70 and a standard deviation of 8. Keep in mind, this is not recalculated every year. The actual average in a given year has tended to creep up and is usually somewhere in the low-to-mid 70s. Similarly, at its creation, the USMLE average was 200 with a std of 20. Now the average is >220.

It’s tempting to assume that this two-digit score is a “raw” score (i.e. that it represents a percentage of correct answers); it does not. The relationship between raw percentage and the NBME two-digit score is not publicized nor is it released. You will not see it. Your school may be able to ask for this information in order to use it for grading purposes (see this interesting post), but as a student you will never know the calculus.

It’s also sometimes thought to reflect a percentile—it does not. Your score report does indeed have a percentile breakdown for a given score if you read through it (sample). If you want to have a general idea in advance (or for estimating your USMLE percentiles, which are not released), you can always use this handy percentile calculator to see. When your NBME score is >92-93, your percentile is generally greater than 99% (with the exception of the psychiatry shelf, which seems to have a significantly larger number of 90s raw scores). So those individuals who manage to get a 99 on their shelf exams have no way of knowing exactly how much they outperformed their peers, nor is it publicized how much better you have to do in terms of percentage correct in order to get these scores.

It’s also worth noting again that the two-digit NBME score has nothing to do with the two-digit USMLE score, which also has nothing to do with percentiles. In fact, later this year the two-digit USMLE score is going the way of the dodo (good riddance).

The score of 70 +/- 8 is already scaled to approximate a normal distribution (although at the lower margin it’s probably skewed with a left tail).

As always, keep in mind this point that I’ve made previously: not all schools use shelf exams the same way. Though the shelf averages were determined by using first-time test-takers using the NBME exam as a final exam, schools often use the shelf as extra-credit, a pass-fail test, a true graded final exam, or as a yay/nay determinant of honors. Your performance can also hinge a bit depending on what part of the year you take the exam, as scores tend to go up a small amount over the course of the year, presumably due to the accumulation of generally useful clinical knowledge (aka the “there was a lot of medicine on that exam” phenomenon). For this reason, your NBME shelf exam percentile is not as reflective of your performance compared to your peers as it would be for the USMLE exams, when all takers have the same amount of skin in the game.


Brant 06.26.13 Reply

Nice article, good job summarizing all the circumstantial explanations that come up when you google this. The links are very helpful too, with the exception of the Normal Percentile Calculator (would everyone quit trying to make me remember stats already, geez!)

kara 08.06.14 Reply

Hi Ben, thanks again for putting up the info for shelf. I’m wondering if you have any inputs/advice for shelf neurology? thanks. K

Ben 08.06.14 Reply

There’s a bit about the neurology shelf in the comments of my clinical (third-year) shelf post.

kara 08.06.14 Reply

thanks Ben!!

Leave a Reply