Step 1 Has No Past Papers: How to Study for It, and How to Know You're Ready (2026)
In many countries, you prepare for the national medical exam with past papers. The questions are published after each sitting, and students work through years of them. The USMLE publishes none, and rebuilding its questions from memory, or using a course that does, can cancel your scores and bar you from the exam. Passing also tells you less than you might expect: Step 1 is pass/fail, and if you pass, your score report gives no feedback at all. In 2025, a quarter of first-time examinees from medical schools outside the US failed. For doctors and students trained abroad, the practical questions are therefore how to study without past papers, how long it takes, and how to know before the day that you are ready. This article answers them from the USMLE's own rules, NBME's guidance on its self-assessments and the research on how people learn, as things stand in November 2026.
1. What you are preparing for
Step 1 tests whether you understand and can apply the basic sciences, with a special emphasis on the principles and mechanisms behind health, disease and treatment. Our article on Step 1 itself covers registration, test centres and the content outline. For planning your study, these are the facts that matter:
| Step 1 | As of November 2026 |
|---|---|
| Format since 14 May 2026 | About 280 questions in fourteen 30-minute blocks of up to 20 questions, in one 8-hour day: about 90 seconds a question |
| Result | Pass or fail only, since 26 January 2022. The passing standard is 196 on the old three-digit scale, which is no longer reported |
| Feedback | None if you pass. If you fail, a chart of each content area against a low pass |
| First-time pass rate, 2025 | 93% for students at US MD schools; 75% for examinees from schools outside the US (22,066 first-time takers) |
| Repeat takers from schools outside the US, 2025 | 54% passed (3,594 examinees) |
| Attempts | Four in total for each Step. A failure stays on your transcript, which residency programs see |
Two things follow. First, the target is a clear pass, not a high score: no program will ever see a number for a Step 1 taken since 26 January 2022. Second, the cost of failing is high and lasting. Barely half of those who retake it pass, and the failure itself stays on the record that programs read. The best plan is to sit Step 1 once, when the evidence says you are ready.
2. No past papers, and why "recalls" are dangerous
The USMLE does not release its exams. Its Bulletin of Information lists as irregular behavior seeking or obtaining unauthorized access to examination materials, reproducing them "through memorization", and discussing test questions with other examinees, potential examinees or "formal or informal" preparation groups at any time before, during or after an exam. It also says it is unlawful for a preparation company to use or distribute content from recent test takers. If you use a course or service that does, your registration or testing may be cancelled, your scores withheld or cancelled, and further sanctions may follow. A finding of irregular behavior is noted permanently on your USMLE transcript, and you may be barred from future exams.
In practice, this means that files sold or shared as "recalls" or "real exam questions" are not a shortcut. They put the exam itself at risk. What the USMLE offers instead is the following:
- The content outline. The Bulletin calls it a complete description of the concepts and principles you are responsible for, and the official advice is a "general, thorough review" of the content the exam covers.
- More than 100 free sample questions, on usmle.org as a PDF and in the new 30-minute block software.
- NBME self-assessments (CBSSA), which are sold separately and estimate your chance of passing (see section 6).
- A practice session at Prometric, for a fee once you hold a scheduling permit. It uses the free sample questions and shows you the test centre.
The sample questions also show what Step 1 means by knowing a subject. For pharmacology, the USMLE says the primary focus is on how drugs work rather than on choosing treatments, drug interactions, adverse effects or contraindications.
3. What the research says works
Without past papers, the question becomes how to learn a very large body of mechanisms so that you can use them on questions you have never seen. Learning research has tested that directly, including in medical students and residents.
| Technique | What the studies found |
|---|---|
| Testing yourself (retrieval practice) | Residents who took repeated tests with feedback on one topic remembered 39% of it six months later, against 26% for a topic they reviewed by reading (Larsen and colleagues, 2009, randomised). In first-year medical students, testing beat studying at six months, 36% to 20% (2013) |
| Spacing your reviews | In a trial of 724 urology residents, diagnosis questions repeated in spaced cycles over 16 weeks produced lasting gains of 15.2%, against 3.4% for the same questions done together in web modules (Kerfoot and colleagues, 2010). The best gap between reviews grows with the time left before the test: weeks, not days, for a test months away (Cepeda and colleagues, 2008) |
| Reading explanations | On the same questions seen again, explanations added little over being shown the right answer. On new questions that needed the same idea, students who had read explanations scored 45%, against 30% for those shown only the answer (Butler and colleagues, 2013) |
| Rereading and highlighting | Rated low utility in a review of ten techniques, which rated practice testing and spaced practice high (Dunlosky and colleagues, 2013). Students who only reread predicted they would remember more, and remembered least (Roediger and Karpicke, 2006) |
| Mixing topics (interleaving) | Mixed results in medicine. Mixed practice helped students read new ECGs in one study, while another found blocked practice better two to four weeks later. A meta-analysis found mixing helps most when the things to tell apart look alike (Brunmair and Richter, 2019) |
The Roediger and Karpicke finding matters for planning. Rereading notes feels productive because the material looks familiar, but familiarity is not recall. Answering questions feels harder and slower, and that difficulty is part of why it works.
4. Using practice questions well
Practice questions are where those techniques meet the exam: each one is a retrieval test with feedback. Studies of Step 1 itself point the same way, though all of them are surveys or records from single US schools, so they show associations rather than cause and effect:
- At Washington University in St. Louis, students who sat Step 1 when it still had three-digit scores had done 3,870 practice questions on average. After allowing for their earlier grades, admission test scores and test anxiety, each extra 1,000 questions went with about 2 more points (Deng and colleagues, 2015).
- At a public US school, starting before the protected study period and doing more questions were both linked to higher scores; the average was 3,597 questions (Burk-Rafel and colleagues, 2017).
- Since pass/fail, a school in South Carolina that could count questions from its learning platform found that students who did more of them were more likely to take and pass Step 1 on schedule. Its flashcards and videos showed no link with practice-exam scores (Sherman and colleagues, 2026).
How you use them matters as much as how many you do.
- Time them like the exam. Twenty questions in 30 minutes is the rhythm of every Step 1 block since May 2026. Pace problems are easier to fix in practice than on the day.
- Mix subjects once you have covered them. On the exam you are not told the subject, and the hard part is often deciding which mechanism a case is about. While you are still learning a system, doing its questions together is reasonable; after that, mixed blocks are closer to the exam. This order is our suggestion: the studies above support mixing confusable topics, not a particular schedule.
- Read every option's explanation. For each wrong option, ask when it would be the right answer. That turns one question into four or five. Explanations of the right answer helped on new questions in the Butler study; going through the wrong options as well is our suggestion, not something a study has tested on its own.
- Review mistakes after a gap. Come back to topics you missed after weeks, not just the next day, and keep some unseen questions for the last weeks rather than finishing them early.
- Do not count a second pass as progress. Questions you have seen before measure memory of the question, not readiness. The same is true of repeating a self-assessment (section 6).
Here is a question from USMLE BANK, written for Step 1. Several options describe real effects of digoxin or of other rate-control drugs. Only one explains this patient's numbers, and finding it takes physiology as well as pharmacology.
A 74-year-old man with permanent atrial fibrillation comes to the office for follow-up. He cannot take beta blockers because of severe reactive airway disease, and diltiazem was stopped because of marked constipation and leg edema. For the past 2 months he has taken digoxin alone for rate control. At rest his heart rate is 78/min and irregular, but when he climbs a flight of stairs his heart rate rises to 150/min and he becomes short of breath. Serum digoxin concentration is within the therapeutic range, and serum potassium is normal.
Which of the following best explains the difference between the resting and exertional heart rates in this patient?
- Decreased inhibition of the sodium-potassium ATPase at higher heart rates
- Direct blockade of L-type calcium channels in the atrioventricular node by digoxin
- Enhancement of vagal tone at the atrioventricular node that is overridden by sympathetic activation during exercise
- Increased ventricular contractility that raises stroke volume during exercise
- Shortening of the atrial refractory period that occurs with exercise
Show the answer and explanation
Answer: C
His digoxin level and potassium are normal, and his rate is controlled at rest but not on the stairs. Digoxin slows the ventricular rate in atrial fibrillation mainly by raising vagal tone: at the atrioventricular node, acetylcholine acting on M2 receptors slows conduction and lengthens the refractory period. During exercise, sympathetic activation through beta-1 receptors overrides that effect, so digoxin controls the rate at rest but poorly on exertion. Three of the wrong options are right answers to other questions. Direct blockade of nodal L-type calcium channels (B) is how verapamil and diltiazem control the rate, which is why they still work during exercise. Increased contractility (D) is why digoxin can relieve symptoms in heart failure with reduced ejection fraction, but it does not change conduction through the node. A shorter atrial refractory period (E) is a digoxin effect that favours atrial arrhythmias, but the ventricular rate is set by the node. How strongly digoxin inhibits the pump (A) depends on its level and on potassium, not on the heart rate.
Source: USMLE BANK / Cardiovascular System #1 Q13
5. How long it takes
The USMLE gives no figure, and we found no study that measures it for doctors trained abroad. What exists is research on US students, who typically set aside a "dedicated" period of 4 to 8 weeks. Read it carefully: those weeks come straight after the basic-science years, and the USMLE notes that most US students take Step 1 at the end of their second year.
| Study | US students' dedicated study period |
|---|---|
| Four schools, 2019 (Tackett and colleagues, 2022) | A median of 7 weeks at 70 hours a week. 77% felt burned out and 72% said their quality of life got worse |
| One school, 2014–15 (Burk-Rafel and colleagues, 2017) | 35 days at 11 hours a day, but 77% had started studying in the preclinical years |
| University of Michigan, pass/fail era (Wang and colleagues, 2024) | 8 weeks was the most common choice; 41% lengthened their period |
More weeks did not mean better results. At Tulane, students who prepared for more than 40 days did not score higher (Kumar and colleagues, 2015), and at another school the number of days studied did not correlate with the score, while a practice exam taken before the study period did (Giordano and colleagues, 2016). In the pass/fail era, students at Albert Einstein College of Medicine who passed on schedule had studied for a median of 42 days; those who needed more time had studied for 62, and the reason they gave most often for delaying was burnout or mental health (Spier and colleagues, 2026). In these studies, a longer study period looked more like a sign of a weaker start than a route to a better result.
For international graduates, those weeks are the wrong yardstick. You may have finished the basic sciences years ago, learned them in another language, or be studying while you work. We found no study of how long that takes, and we will not invent a number. What the evidence does support is planning by checkpoints rather than by the calendar:
- Measure where you start. NBME suggests a self-assessment early on as a baseline. The distance between that score and the low-pass range (section 6) is a better guide to the time you need than someone else's schedule.
- Start practice questions early, during the foundation phase rather than after it. In the Burk-Rafel study, starting before the protected period went with higher scores.
- Plan hours you can keep. The US dedicated period is a short sprint, and burnout was the main reason students delayed. Our view: if you are working, a longer plan at a pace you can keep is more realistic than copying a 70-hour week, as long as your self-assessment scores keep rising.
6. How to know you are ready
Because a pass gives no feedback, the best check you have before the day is the NBME's Comprehensive Basic Science Self-Assessment (CBSSA). It comes from NBME, one of the two organisations behind the USMLE, and its report estimates your chance of passing. Seven forms were on sale in November 2026, from $62 each. A form has four sections of 50 questions, with 1 hour 15 minutes for each section in standard-paced mode: the same 90 seconds a question as the exam. A self-paced mode gives four times as long.
NBME's guide to reading the report (September 2026) asks you to look at three things together:
- Your score and its likely range. The report gives an equated percent correct score with a range of about 4 points either way, where your score would fall two times in three if you sat it again without learning or forgetting anything.
- Where that range sits against the Step 1 low-pass range. If your whole range is above it, you are likely ready. If it overlaps the low-pass range, NBME says to take extra caution, and if any of it is below, it strongly recommends more preparation. Two or more forms partly above the low-pass range may reassure more than one.
- Your estimated probability of passing, from 1% to 99%, modelled on examinees who sat Step 1 for the first time within a week of the self-assessment.
The guide's own warnings are worth repeating. A 90% probability means that about one in ten people with the same performance would be expected not to pass. The estimate is less reliable if you take the form more than a week before the exam, use the self-paced mode, look things up during it, or take a form whose questions you have already seen. Sit each form once, in one sitting, under exam conditions, and treat the results as a trend rather than a verdict.
NBME's own researchers checked how well this works since pass/fail, using 23,271 students from US and Canadian schools who sat Step 1 between February 2022 and March 2023 (Fan and colleagues, 2025). In that period the low-pass range sat between 62 and 68 on the CBSSA's percent-correct scale.
| Last CBSSA | What happened on Step 1 |
|---|---|
| Score of 62, within two weeks of the exam | Estimated probability of passing about 91% |
| Score of 52, within two weeks | About 50% |
| Taken 1–2 weeks before the exam | The CBSSA's pass or fail matched Step 1's for 88% of students |
| Taken 5–6 weeks before | It matched for only 50% |
The low agreement at five to six weeks comes mostly from students who were below the cut on the CBSSA and passed Step 1 anyway, presumably after more weeks of study. So a self-assessment taken five weeks or more before the exam says little about the result; the one that counts is the last one, close to the day. The students in the study were all from US and Canadian schools, and we found no comparable study of international examinees, which is one more reason to want a margin above the low-pass range rather than a score inside it.
The free USMLE sample questions are worth sitting under exam conditions too, but they give a percentage correct, not a prediction.
7. Booking the date
When you apply, you choose a three-month eligibility period, and seats at Prometric go to whoever books first. Moving a booking is free until 46 days before the exam and then costs more the closer you are; our article on Step 1 lists the fees. One way to use this: book a date far enough ahead to commit to, and move it early if your self-assessments say you will not be ready.
Do not sit Step 1 "to see what it is like". The Bulletin says that taking a Step to familiarise yourself with the format, or for any reason other than to pass, is prohibited and may restrict your future access to the USMLE, and that you should not test if you are not prepared. Performance that suggests a lack of readiness can itself lead to a review of your eligibility.
Step 1 and Step 2 CK can be taken in either order. For ECFMG Certification, though, the exams must all be passed within seven years of the first one you pass, and most state boards set a similar clock for all three Steps. If you are still a student and Step 2 CK or residency is years away, an early pass starts that clock.
8. If you do not pass
A failing score report shows, for each content area, whether you were below, about the same as or above a low pass. It is tempting to study only the weak areas. The USMLE advises against it: the exams are highly integrative, remediation that targets only the weakest areas is unlikely to work best, and in most cases a failure is best remedied by reviewing everything, weighted by how much of the exam each area makes up and how you did in it.
The attempt rules shape the timing. You can take Step 1 at most three times in 12 months; a fourth attempt must be at least 12 months after the first and six months after the most recent, and four unsuccessful attempts at any Step end your eligibility for all of them. With a repeat pass rate of 54% for international examinees in 2025, a retake deserves the same checks as the first attempt: a new plan, new self-assessment forms, and a date set by the results.
9. A plan in four phases
This is our summary of the evidence above, not an official plan. It is organised by what you need to be able to do, because the right number of weeks depends on where you start.
- Phase 1, foundations: go through the content outline system by system, doing questions on each system as you finish it. Read every explanation. Note the mechanisms you could not have explained without the book.
- Phase 2, mixed and timed: switch to mixed blocks of 20 in 30 minutes. Keep a list of missed concepts and come back to them after a few weeks.
- Phase 3, checkpoints: sit the free sample questions and then CBSSA forms under exam conditions, spaced out, each form only once. Let the trend decide the date.
- Phase 4, the last weeks: keep doing new mixed blocks rather than rereading. Take a last CBSSA within a week or two of the exam, and move the date if your likely range still overlaps the low-pass range.
Frequently asked questions
How long does it take to prepare for USMLE Step 1?
There is no official figure. US students usually set aside 4 to 8 weeks of full-time study straight after their basic-science years, and longer study periods did not go with better results. No study has measured it for doctors trained abroad, so set your date by NBME self-assessments taken under exam conditions rather than by a number of weeks.
Are there past papers for USMLE Step 1?
No. The USMLE does not release its exams. Rebuilding questions from memory, sharing them, or using a course that uses real exam content is irregular behavior: it can cancel your scores, is noted permanently on your transcript and can bar you from the exam. The official material is the content outline, more than 100 free sample questions, NBME self-assessments and a practice session at Prometric.
How many practice questions should I do for Step 1?
No study sets a number. In single-school US studies, students did about 3,600 to 3,900 questions on average, and more questions went with better results, but these are associations, not proof. How you use them matters too: timed, mixed blocks of 20 in 30 minutes, reading every option's explanation, and coming back to missed topics after a gap of weeks.
What CBSSA score means I am ready for Step 1?
NBME says you are likely ready when your whole likely score range is above the Step 1 low-pass range. In NBME's study of 2022–23, that range sat at 62 to 68 on the CBSSA's percent-correct scale, and a score of 62 within two weeks of the exam meant about a 91% estimated chance of passing. Even a 90% estimate means about one in ten would not pass, so aim for a margin.
Can I take Step 1 just to see what it is like?
No. The USMLE Bulletin says that taking a Step to familiarise yourself with the format, or for any reason other than to pass, is prohibited and may restrict your future access to the exams. A failure also stays on your transcript, and only 54% of repeat takers from schools outside the US passed in 2025.
The exam: USMLE — Step 1; USMLE — 2027 Bulletin of Information (Preparing for the USMLE; Scoring and Score Reporting; Irregular Behavior); USMLE — Common Questions (Practice Materials, Score Reports, Attempt Limit); USMLE — Step 1 Sample Test Questions; USMLE — Performance Data (2025)
Self-assessment: NBME — Comprehensive Basic Science Self-Assessment; NBME — CBSSA and CBSE: Gauging Readiness for USMLE Step 1 (September 2026); NBME — CBSSA sample Examinee Performance Report
Learning research: Roediger and Karpicke, Psychological Science 2006;17:249; Larsen, Butler and Roediger, Medical Education 2009;43:1174 and 2013;47:674; Kerfoot and colleagues, Journal of the American College of Surgeons 2010;211:331; Cepeda and colleagues, Psychological Science 2008;19:1095; Butler, Godbole and Marsh, Journal of Educational Psychology 2013;105:290; Dunlosky and colleagues, Psychological Science in the Public Interest 2013;14:4; Hatala, Brooks and Norman, Advances in Health Sciences Education 2003;8:17; Monteiro and colleagues, Perspectives on Medical Education 2017;6:237; Brunmair and Richter, Psychological Bulletin 2019;145:1029
Step 1 studies: Deng, Gluckstein and Larsen, Perspectives on Medical Education 2015;4:308; Burk-Rafel, Santen and Purkiss, Academic Medicine 2017;92(11S):S67; Sherman, Stomberger and Porter-Stransky, BMC Medical Education 2026;26:1453; Kumar and colleagues, Postgraduate Medical Journal 2015;91:257; Giordano and colleagues, Cureus 2016;8:e769; Tackett and colleagues, BMC Medical Education 2022;22:16; Wang and colleagues, Medical Science Educator 2024;34:807; Spier and colleagues, Medical Science Educator 2026;36:611; Fan and colleagues, Medical Science Educator 2025;35:149
Practice that explains every option.
USMLE BANK is writing original Step 1 questions explained from the clues to the diagnosis to the mechanism, going through every answer choice. It is not open yet: three sample questions are on the home page, and you can leave your email to hear when it opens.
Try the sample questions