# How to Review UWorld Incorrect Questions

> How should I review UWorld incorrect questions?

A repeatable routine for wrong answers: an error-type taxonomy, a per-question checklist, and the block-level patterns that tell you what to change next.

Review every incorrect question in three steps. First, diagnose the error type (knowledge gap, misread vignette, faulty elimination, time pressure, or second-guess reversal) from what you were thinking when you committed to an answer. Second, fix the underlying gap rather than memorizing that item's answer. Third, re-test the fix on fresh questions within a week.

## How should I review UWorld incorrect questions?

An incorrect answer carries two pieces of information: the fact you did not know, and the reasoning that led you to a distractor. Most students harvest only the first. The explanation teaches the content; nothing in it tells you why **you** picked choice C.

Effective review separates those layers: recall what you were thinking, name the error type, then decide on the fix.

The third step is the one almost everyone skips: a gap closes only when you retrieve the fact cold, days later, from a different stem. Schedule that re-test into your [Step 1 study schedule](/blog/usmle-step-1-study-schedule).

> **If you only remember one thing:** the explanation tells you the right answer; only your reasoning tells you why you were wrong. Write down the second one.

## Error types: how to tell which mistake you actually made

Five error types account for nearly every wrong answer, each identifiable from your own recollection rather than the explanation text.

| Error Type | What It Means | What To Do Next |
| --- | --- | --- |
| **Knowledge gap** | You did not know the fact or mechanism tested; more time would not have produced it. | Learn the concept at mechanism level, write one line linking cause to consequence, schedule retrieval. |
| **Misread vignette** | You knew the content but missed a decisive detail: an age, a time course, a lab value, *initial* versus *most accurate*. | Slow the first read, and re-read the stem's last sentence before the choices. |
| **Faulty elimination** | You narrowed to two choices, then chose on familiarity instead of a discriminating feature. | Write the one feature separating the final two, then drill that pair. |
| **Time-pressure mistake** | The clock forced an answer before the reasoning finished; you can finish it now unprompted. | Fix pacing, not content: about 90 seconds per item, and flag rather than guess-and-abandon. |
| **Second-guess reversal** | You had the right answer selected and talked yourself out of it on a distractor's appeal. | Log what the distractor promised; change an answer only for a newly noticed detail. |

## The step-by-step workflow for one incorrect question

Budget three to five minutes per incorrect item, less for a diagnosed misread, and end the review when you have written one line, not when the explanation ends. Run this on every wrong answer:

1. **Re-answer the stem before opening the explanation**, untimed. Getting it right now separates a timing error from a knowledge gap.
2. **Name the error type** from the table and record it -- the type, not a feeling.
3. **Read the explanation for the correct choice**, then the distractor paragraph for your choice. Skip the rest.
4. **Write one sentence in your own words**: stem trigger -> concept -> answer.
5. **Add the discriminating feature** separating your choice from the right one.
6. **Tag the item twice** -- one organ system, one mechanism label: drug class, enzyme, or receptor.
7. **Convert the sentence into a retrieval cue** -- a flashcard or self-quiz prompt hiding the answer.
8. **Schedule the re-test**: fresh questions on that tag within three to seven days, never the same item.

Never write down the explanation verbatim, the full vignette, a question ID, or a fact you already knew.

## What your error pattern across a block tells you to change

Tag every incorrect item, and the block pattern teaches more than any single question.

| Dominant pattern | What it means | Change to make |
| --- | --- | --- |
| Errors cluster in one organ system | A content gap, not a test-taking problem | Re-study the system, then run a targeted set before the next mixed block |
| Errors cluster on drug mechanisms and adverse effects | [Pharmacology](/topics/pharmacology) is the weak axis, not any one organ | Drill drug classes by mechanism and keep them in daily retrieval |
| Errors scattered across systems, all feeling like "I knew that" | Stem processing, not content | Do a block untimed, then re-introduce the clock as accuracy recovers |
| Errors concentrated in the last ten questions of a block | Pacing and stamina | Practice full 40-question blocks rather than more content review |

## When passive review stops working and active retrieval takes over

Re-reading an explanation produces a strong feeling of knowing and weak retention. In classic experiments, repeated testing after learning produced large one-week recall gains where repeated study produced almost none ([Karpicke & Roediger, 2008](https://pubmed.ncbi.nlm.nih.gov/18276894/)).

Three signals mean review has gone passive: you recognize the answer but cannot produce it unprompted, your notes grow while your percentage stays flat, or you miss the same concept twice weeks apart. All three take the same fix -- turn the note into a question and space the repetitions, which a [spaced-repetition flashcard system](/blog/spaced-repetition-usmle-flashcards) automates.

## Worked examples: two missed questions, start to finish

A stem describes a patient started on lisinopril six weeks earlier with a persistent dry cough, no fever or wheeze, asking for the mechanism. You chose direct airway irritation; the answer is decreased degradation of bradykinin.

Re-answered untimed you still miss it: a knowledge gap, not timing. ACE is also kininase II, so inhibiting it lets bradykinin accumulate, producing cough and, less often, angioedema. The line you write: "ACE inhibitor -> ↑ bradykinin -> dry cough and angioedema; switch to an ARB, which spares bradykinin." Tag it cardiovascular plus ACE inhibitors, and re-test on a fresh set four days later.

A second pattern: two months of amlodipine, new bilateral ankle edema, normal jugular venous pressure, clear lungs. The answer is arteriolar dilation raising capillary hydrostatic pressure, not volume overload -- a misread if you skipped the drug list.

## Common review mistakes that waste the block

- **Reviewing only the incorrects.** A flagged item you guessed is an unlogged gap that will not stay lucky twice.
- **Reviewing every item at the same depth.** A misread vignette needs thirty seconds; an unfamiliar mechanism needs the full workflow.
- **Batching review to the weekend.** After 120 items you cannot recall what you were thinking, which destroys the diagnosis.

## Turning the weak-topic list into targeted practice

A tagged weak-topic list only helps if it changes what you practice next. A targeted set on that tag makes you produce the answer from an unfamiliar stem -- the only evidence a gap has closed. Run a set on each tag, then send whatever you still miss back to step one.

## Sources

- [USMLE Step 1 exam format and content outline](https://www.usmle.org/step-exams/step-1)
- [Karpicke & Roediger, the critical importance of retrieval for learning](https://pubmed.ncbi.nlm.nih.gov/18276894/)

[Enter your weak topics in StepGenie and get targeted questions immediately](https://dashboard.stepgenie.app/sign-up), and put every fix back under a stem the same week.

## Frequently asked questions

### How long should you spend reviewing each incorrect question?

Three to five minutes per incorrect question is the working budget for an item you genuinely did not know. Anything faster usually means you read the explanation without diagnosing why you chose a distractor, and anything much longer means you are copying it rather than condensing it. A miss you have already diagnosed as a simple misread closes faster. End the review once you have written one sentence in your own words and tagged the item.

### Should you review the questions you answered correctly?

Review the correct answers you flagged, guessed, or narrowed to two choices, and skip the rest. A lucky guess is an unlogged knowledge gap that will not stay lucky on test day, so it deserves the same one-line note as a miss. Items you answered confidently and for the right reason need no review at all and reading them wastes block time.

### What belongs in a USMLE error log, and what does not?

An error log should hold the error type, one sentence in your own words linking the stem trigger to the concept and answer, the feature that separates your choice from the correct one, and two tags: organ system and mechanism. It should never hold the explanation copied verbatim, the full vignette, or a question ID treated as if it were content.

### Is it better to review incorrect questions right away or after the block?

Review at the end of the block, on the same day, never days later. Reviewing mid-block breaks the timed conditions the block is meant to simulate, but waiting until the weekend means you can no longer recall what you were thinking when you answered, and the error type is the one thing only you can recover.

### Does redoing the same UWorld questions actually help?

Redoing an item you have already reviewed mostly tests your memory of that item, not the concept, because you recognize the stem rather than reasoning through it. Re-test a fixed gap on fresh questions carrying the same tag, three to seven days later. If fresh questions on that tag are not available, quiz yourself from the concept side instead.

---

Canonical page: [How to Review UWorld Incorrect Questions](https://www.stepgenie.app/blog/how-to-review-uworld-incorrect-questions)
Topic hub: [Study Strategy](https://www.stepgenie.app/topics/pharmacology)