0.3 Scoring, the Passing Standard & Score Reports
Key Takeaways
- ABIM reports scores on a standardized 200-to-800 scale with a mean of 500 for traditional assessments.
- The passing standard is derived by the Angoff method, in which physician experts judge each item, and is then set by the Specialty Board and periodically reviewed.
- Scoring is criterion-referenced, so results depend on a fixed competence standard rather than on how the rest of the cohort performed.
- Approximately 35 unscored pretest items are embedded in each form and cannot be identified, so every question must be answered as though it counts.
- Score reports include an overall scaled score plus content-area subscores and standard error information; the subscores are diagnostic only and do not determine pass or fail.
The Scale
ABIM reports performance on the Internal Medicine Certification Exam using a standardized score scale running from 200 to 800, with a mean of 500. A single scaled score determines the outcome. There is no honors tier, no percentile requirement, and no partial credit.
ABIM does not publish the numeric passing standard on its scoring page. What it does publish is the method: the standard is derived using the Angoff method, in which panels of physician experts judge, item by item, whether a minimally qualified internist would answer correctly. Those judgments are combined into a recommended content-based standard, and the Internal Medicine Specialty Board then sets the actual standard, taking into account changes in the field and characteristics of the candidate population. The standard is periodically reviewed but does not move from administration to administration in response to cohort performance.
A scaled score of 366 is the value widely reported for the Internal Medicine Certification Exam by board-review and preparation resources, and it is a reasonable working figure. Treat it as a reported convention rather than an ABIM-published constant, and rely on the mechanism — a fixed, expert-derived standard — rather than on a number when reasoning about the exam.
Criterion-Referenced, Not Curved
This is the single most misunderstood feature of the exam, and it has direct behavioral consequences.
| Norm-referenced (curved) | Criterion-referenced (ABIM) | |
|---|---|---|
| What you are compared to | Other candidates on your form | A fixed standard of competence |
| Effect of a strong cohort | Harder to pass | No effect |
| Effect of a weak cohort | Easier to pass | No effect |
| Quota on passers | Implicit | None |
Because the standard is fixed, every candidate on your exam date could pass, or none could. There is nothing to be gained from speculating about how others are doing, and nothing is lost by helping a co-resident study.
Why There Is No "Number Right to Pass"
Candidates routinely ask how many questions they must answer correctly. ABIM does not publish such a figure, and the reason is structural rather than evasive: raw-to-scaled conversion varies by exam form. Different forms are assembled from different item pools and differ slightly in difficulty. Scaled scoring exists precisely to make a 366 on an easier form equivalent in meaning to a 366 on a harder one, which requires the raw threshold to move between forms.
Any specific percentage circulating in review-course folklore is therefore an estimate applied to an unknown form, not a published standard.
Pretest Items
Roughly 35 of the up-to-240 questions on your form are new items being pretested and do not contribute to your score. You cannot identify them. They are not clustered at the end, not visually marked, and not necessarily harder than scored items — a question that feels bizarre is at least as likely to be a poorly worded scored item as a pretest item.
The operational rule is simple: answer every question with full effort, and never leave one blank. There is no penalty for a wrong answer, so an unanswered question is strictly worse than a guess.
The Score Report
A passing or failing result is accompanied by a performance profile broken out by blueprint content area — cardiology, pulmonary, gastroenterology, endocrinology, infectious disease, hematology, oncology, rheumatology, nephrology, neurology, psychiatry, dermatology, geriatrics and the rest.
Three things to understand about that profile:
- It does not determine pass or fail. There is no per-domain minimum. A single overall scaled score decides the outcome. You can be at the bottom of the distribution in dermatology and still pass comfortably.
- Its precision is limited. A 1% category contributes roughly two scored questions. A performance band built on two items carries enormous measurement error and should not drive a re-study plan.
- It is genuinely useful for re-takers in the large categories. A weak band in a 9–14% category rests on 18–29 items and is a real signal.
ABIM does not disclose which specific questions you answered incorrectly, and it does not release raw scores.
Rescores
A candidate may request a rescore for a fee of $250. A rescore is a verification that the scoring process was applied correctly to your responses; it is not a re-reading of your answers by a human expert and it does not involve re-judging any item. Rescores essentially never change an outcome, because the underlying process is automated and audited. Request one only if you have a specific reason to believe an administrative error occurred.
Retakes
A failing result is not the end of board eligibility. Candidates may re-register for a subsequent administration, subject to the seven-year eligibility window, and each attempt carries the full examination fee. Because the exam is administered only in August, a failed attempt costs a full year — which is the strongest practical argument for sitting early in your eligibility window rather than late.
Using the Standard Strategically
The combination of a fixed standard, no per-domain minimum, and no penalty for guessing produces a clear optimization:
- Do not sacrifice breadth for depth in a favorite subspecialty. Extra mastery in cardiology beyond the level needed to answer typical items adds nothing once those items are already correct, while an unopened dermatology chapter leaves cheap points on the table.
- Protect the mid-weight categories. Hematology, oncology and nephrology at 6% each total 18% of the exam — more than cardiovascular disease — and are commonly under-studied relative to that.
- Answer everything. With no wrong-answer penalty, a 25% blind guess has positive expected value and an omission has none.
A candidate scores 358 on the Internal Medicine Certification Exam and learns that the mean scaled score for her administration was 512, the highest in five years. She asks whether the unusually strong cohort raised the bar against her. What is the correct response?
During the final session of the exam, a candidate encounters a question about an obscure syndrome she has never seen. Two of the four options are clearly wrong. She has 40 minutes and 12 questions remaining. What is the optimal action?