This is for internal medicine physicians and residents using Rosh Review's Internal Medicine Certification Qbank for the ABIM exam who want to read its analytics—especially its Probability of Passing and Projected Score—without over-trusting them. Its strength is a large, well-explained, blueprint-mapped bank with unusually transparent performance metrics. Its principal limitation is that those headline predictors are computed from questions you have chosen and often reviewed, so they can read high before your performance on unseen, mixed, timed material has caught up.
What Rosh Review and Blueprint Prep offer for ABIM right now
One clarification first, because the branding confuses people. Blueprint Prep (formerly Blueprint Test Prep) acquired Rosh Review in 2021. Blueprint's own consumer lines are LSAT, MCAT, USMLE/COMLEX, PA and nursing; it does not sell a separate Blueprint-branded ABIM product. The ABIM Internal Medicine qbank is sold under the Rosh Review name, now running on Blueprint's platform—so "Blueprint Prep ABIM" is, in practice, Rosh Review Internal Medicine.
| Feature (last checked 19 July 2026) | What the vendor states | Confidence |
|---|---|---|
| Owner/brand | Rosh Review, a Blueprint Prep company (acquired 2021; migrated to Blueprint's platform in 2024) | Confirmed |
| Question bank | 3,000 questions written to the ABIM Certification blueprint | Confirmed (product page) |
| Mock exam | 240-question mock mapped to the blueprint (Premium tier) | Confirmed |
| "Adaptive"? | No computer-adaptive engine; custom learner-built quizzes in tutor (untimed) or test (timed) mode | Confirmed absence |
| Teaching | Detailed explanations with images for right and wrong answers; "One Step Further" bonus question; "Rapid Review" summaries | Confirmed |
| Headline analytics | Overall Performance (% correct); Probability of Passing (%); Projected Score; peer answer-choice comparison; category breakdown; score trends | Vendor-reported |
| Pass guarantee | Complete 100% of questions and, if you do not pass, a full refund or a free extension to the next exam | Confirmed |
| Access/price | 1-year Basic ~US$589 / Standard ~US$689 / Premium ~US$799; 30- and 90-day terms exist but prices not shown—verify | Vendor-reported; some UNCONFIRMED |
| CME | Up to 100 AMA PRA Category 1 Credits (Standard/Premium) | Confirmed |
| ABIM components | Initial certification, MOC and LKA; a separate IM-ITE qbank exists | Confirmed |
The one word to strike from your mental model is "adaptive." Rosh does not select questions by algorithm; you build the quizzes. That matters for reading the analytics, because every number below is computed on a sample you curated.
The ABIM exam you are actually preparing for
ABIM initial certification is up to 240 single-best-answer questions in four sessions of up to 60, across roughly ten hours at Pearson VUE, about 35 items unscored, blueprint-weighted as follows:
| Medical content category | Weight |
|---|---|
| Cardiovascular Disease | 14% |
| Endocrinology, Diabetes and Metabolism | 9% |
| Gastroenterology | 9% |
| Infectious Disease | 9% |
| Pulmonary Disease | 9% |
| Rheumatology and Orthopedics | 9% |
| Hematology, Nephrology and Urology, Medical Oncology | 6% each |
| Neurology, Psychiatry | 4% each |
| Dermatology, Obstetrics and Gynecology, Geriatric Syndromes | 3% each |
| Allergy and Immunology, Miscellaneous | 2% each |
| Ophthalmology, Otolaryngology and Dental Medicine | 1% each |
Cross-content topics (critical care, prevention, epidemiology, ethics, nutrition, palliative care, patient safety, substance misuse) run through the categories. The format and weighting are official ABIM facts; Rosh's Projected Score is a vendor model built on your practice data, and the two must not be conflated.
Define every metric the platform shows you
- Overall Performance is your cumulative percent correct across attempted questions—inflated by any items you have reviewed and re-attempted.
- First-attempt vs repeat accuracy: prioritise first-attempt figures; repeat accuracy drifts toward memory.
- Probability of Passing is Rosh's flagship, shown as a percentage likelihood; the platform recalculates it nightly and states it is more reliable once you have completed at least half the bank (vendor-reported).
- Projected Score estimates your exam score against the target exam set in your profile (vendor-reported).
- Peer comparison shows how your answer choices align with other learners nationally—useful context, not a readiness verdict.
- Category performance breaks accuracy down by subject, which is where the actionable signal lives.
- Score trends track movement over time on the Blueprint platform.
Treat Probability of Passing and Projected Score as vendor-reported models, not measurements. They are computed from questions you selected, in modes you chose (including untimed tutor mode), often after seeing explanations—conditions the real exam will not reproduce.
Selection bias: why your curated sample flatters the numbers
Because you build the quizzes, two biases creep in. You tend to over-practise subjects you enjoy or find tractable and under-practise the ones you dread, so your Overall Performance is weighted toward your comfort zone. And in tutor (untimed) mode you can think without a clock and read explanations mid-set, which lifts accuracy above what a timed, unassisted block would show. The result is that a reassuring Projected Score can sit on top of a sample that is neither blueprint-representative nor exam-like. The fix is not to distrust the platform but to feed it a fair sample—and to corroborate it against unseen, timed, mixed practice, per "Your Q-Bank Percentage Is Not Your Exam Score".
Blueprint audit: check your attempted distribution, not the headline
The Probability of Passing does not know how your attempts map to the blueprint. Tally your attempted questions by category and set them against the ABIM weighting: if cardiovascular (14%) or the 9% blocks are thin because you gravitated to shorter, satisfying topics, your headline number is standing on a skewed base. Rebalance quizzes to the blueprint before you trust any predictor. This distribution check is the blueprint-coverage matrix applied to a bank you steer yourself.
The readiness test: what a credible signal requires
A readiness reading must come from a block that is unseen, timed to ABIM pace (roughly 60 questions per session), mixed across the blueprint, taken with no assistance (test mode, no explanations mid-block, no lookups), and large enough to be stable. Rosh's own 240-question mock, taken cold and once, is a good use of the Premium tier for exactly this. What does not qualify is your cumulative Overall Performance across curated tutor-mode sets, however high the Probability of Passing looks beside it.
Algorithm override rules—here, self-discipline rules
There is no algorithm to override, which shifts the burden onto you. Deliberately schedule the high-weight categories you have been avoiding; force image and data-interpretation items rather than defaulting to text; and force the low-frequency, high-stakes material—ethics, patient safety, calculations, palliative care—that a self-built quiz almost never surfaces unless you ask for it. Use category performance to target the weakest subjects, and switch from tutor to test mode as you approach the exam so your numbers start to resemble exam conditions.
Worked dashboard example: from analytics to next week's quotas
Suppose Rosh shows Probability of Passing 82%, Projected Score comfortably above the line, strong cardiology and endocrine categories, weak nephrology and neurology, and most of your work done in untimed tutor mode. Convert that into quotas—without treating 82% as a verdict:
| Signal on the dashboard | Next week's action | Quota |
|---|---|---|
| Predictors high but built in tutor mode | Move to timed test mode | All blocks timed |
| Weak nephrology and neurology | Targeted category quizzes | 40 items each |
| Attempts skewed off blueprint | Rebalance to weights | Mixed 60-item set |
| No cold mock yet | Take the 240-question mock once, unseen | 1 full mock |
The Probability of Passing is a prompt to test the claim, not a licence to stop.
A worked seven-day plan: Rosh to drill, iatroX to measure
Give Rosh the job it does well—high-volume, well-explained drilling with category diagnostics—and give iatroX the unseen, timed, mixed measurement that a bank you curate cannot supply. No claim is made about either platform's internal scoring.
| Day | Rosh Review (drill) | iatroX (measure) |
|---|---|---|
| Mon | 40 timed test-mode items on weak categories; read explanations after | — |
| Tue | "One Step Further" review; Rapid Reviews on two gaps | — |
| Wed | 40 timed items, blueprint-balanced | — |
| Thu | Image/data and ethics override set | — |
| Fri | Category clean-up on the weakest subject | — |
| Sat | — | 40 fresh, timed, mixed ABIM items in iatroX, no assistance |
| Sun | Compare Rosh predictors with the iatroX result; reset overrides | Read only the iatroX misses |
When the Probability of Passing and your unseen iatroX accuracy agree, you have a real signal. When they diverge, believe the unseen number.
Decision checklist: continue, supplement, switch or stop
Continue if category performance is still improving and you are steadily converting tutor-mode learning into timed test-mode accuracy. Supplement with unseen, timed, mixed practice whenever your Probability of Passing outruns any independent reading—that divergence is the measurable trigger. Switch only if a distribution audit shows the bank has left blueprint areas persistently weak despite targeting. Stop re-attempting items you have reviewed just to lift Overall Performance; that inflates a number without moving readiness. The two-Q-bank rule and the comparison hub cover adding a second bank without duplication.
Bottom line
Rosh Review gives you more analytical transparency than most banks—an explicit Probability of Passing, a Projected Score, peer context and clean category breakdowns—and a large, well-explained, blueprint-mapped bank underneath. The catch is that those predictors are vendor models built on questions you chose, often in untimed mode, so read them as hypotheses to test rather than readiness confirmed. Feed the bank a blueprint-fair, timed sample, take the mock cold, and corroborate against unseen practice; do that and the predictors become genuinely useful.
Frequently asked questions
Is Rosh Review and Blueprint Prep enough for ABIM on its own? As a learning and drilling resource, Rosh's 3,000-question ABIM bank is a strong standalone core with a pass guarantee behind it. Where it falls short of "enough" is independent readiness measurement: its Probability of Passing and Projected Score are computed from questions you selected, often in untimed tutor mode, so they can read high before unseen, timed performance has caught up. Most candidates should drill in Rosh and confirm readiness on fresh, mixed, timed material.
Which ABIM component does Rosh Review and Blueprint Prep not reproduce well? The unassisted, mixed, timed exam experience, unless you actively force it. Rosh defaults to learner-built quizzes and offers an untimed tutor mode, which is the opposite of exam conditions; the 240-question mock is the part of the product that best reproduces the real thing, and it works only if you take it cold. The platform is not computer-adaptive, so it will not automatically balance your practice to the blueprint.
How many Rosh Review and Blueprint Prep questions should I complete per day for ABIM? There is no official figure, and any number is a planning heuristic rather than a vendor claim. A workable pattern for most physicians is 30 to 40 questions a day in timed test mode with explanations reviewed afterwards, which lets you move through a 3,000-item bank over a few months while leaving room for a cold mock and unseen practice near the end.
When should I stop using Rosh Review and Blueprint Prep and move to mixed mocks? When category performance is strong across the blueprint and your Probability of Passing is high but untested against fresh material. That is precisely the moment to take the 240-question mock cold and add unseen, timed, mixed blocks—if the unseen numbers confirm the predictors, you are ready; if they lag, the predictors were flattered by your curated sample.
How should I combine Rosh Review and Blueprint Prep with iatroX without duplicating practice? Split the jobs. Use Rosh for volume drilling, explanations and category diagnostics; use iatroX only for unseen, timed, mixed measurement. Never re-test an item you have already seen in either bank, and when Rosh's Probability of Passing and your iatroX accuracy disagree, trust the unseen figure. Keeping the vendor predictor and the unseen result in separate columns is the discipline that keeps you honest.
Editorial notes and references
Written by Dr Kolawole Tytler, NHS GP and founder of iatroX. Last checked 19 July 2026; question counts, prices, the pass-guarantee terms and the Probability of Passing and Projected Score models are vendor-reported and change—verify current details on the Rosh Review Internal Medicine product and pricing pages. The 30- and 90-day prices were not published at review and should be confirmed directly. Disclosure: iatroX operates a competing ABIM question bank; we confine its role here to the job a self-curated bank cannot do—unseen, timed, mixed measurement—and do not present it as a replacement for Rosh's drilling and explanations. Corrections are welcome via the feedback route on iatrox.com. References: ABIM Internal Medicine certification blueprint and exam information (abim.org); Rosh Review Internal Medicine Certification Qbank (roshreview.com); and "Your Q-Bank Percentage Is Not Your Exam Score" plus the iatroX ABIM bank.