Finishing a first pass through a question bank produces a specific kind of paralysis. You have several weeks left, a bank full of attempted questions, and three plausible options: start again from the beginning, work only through the ones you got wrong, or buy a second bank and start fresh. Candidates agonise over this, and then usually pick the option that feels most industrious rather than the one that works. The right answer depends on two variables, and it is worth being precise about both.
Key takeaways
- The purpose of a first pass is to build a map, not to learn everything, so judge it on coverage rather than score.
- Redoing every question inflates your confidence and measures recognition rather than knowledge.
- Redoing only incorrects is better, but it tests whether you remember the item rather than the concept.
- The best default is a hybrid: fresh unseen questions, plus spaced returns of the concepts you got wrong.
- Which way you lean depends on how far away the exam is and how strong your baseline was.
What a first pass is actually for
Be clear about what you have just done, because it changes how you evaluate it.
A first pass is a reconnaissance exercise. Its purpose is to map the syllabus, expose your weak domains, and identify the operations you are poor at. It is not primarily a learning event, and judging it by the percentage you scored is a category error. A first pass that produced a mediocre score and a complete, honest map of your weaknesses has done its job perfectly.
So before deciding what to do next, extract the map. Which domains are weak. Which have barely been sampled. Which reasoning operations keep failing. Which errors you made while feeling certain. That analysis, not the score, is the output of a first pass, and it is what should drive everything that follows.
Why redoing everything is the weakest option
The most popular choice is a full second pass, and it is popular because it feels rigorous and produces a beautiful rising number.
It is also the least efficient use of your remaining time, for a simple reason. You have seen these questions. When you meet one again, you are not testing your knowledge, you are testing your memory of this item, and those are different things. You will recognise the stem, remember that you chose option C and it was wrong, and select something else. Your accuracy will climb impressively, and the climb will be almost entirely recognition.
Worse, it is actively misleading, because your rising percentage will convince you that you are more prepared than you are. A candidate whose bank says eighty-five per cent on a second pass and who has not touched an unseen question in a month is in a recognition plateau and does not know it.
Why redoing only incorrects is better, but incomplete
The next most popular choice is to redo just the questions you got wrong. This is a genuine improvement, because at least it directs effort at your weaknesses.
But it inherits the same flaw in miniature. Re-attempting the specific item you got wrong tests whether you remember what the correct answer was. It does not test whether you understood the principle, and those come apart constantly. Plenty of candidates can produce the right answer to a question they have already failed and then fail a different question testing the identical rule, because what they retained was the answer and not the reasoning.
The question you want to answer is not "can I now get this item right". It is "can I now get a different question about this principle right", and that requires a different question.
The distinction that resolves it
Separate the item from the concept.
An item is a specific question with a specific stem. A concept is the underlying rule it tests: that this investigation precedes that one, that this drug is contraindicated in that situation, that this threshold triggers that action.
Items are disposable. Once you have seen one, it has spent most of its value. Concepts are what you are actually revising, and the correct thing to carry forward from your first pass is not a list of questions to redo, but a list of concepts you got wrong.
Then the plan writes itself: return to the concepts, in new questions, spaced out over time.
The hybrid default
For most candidates, most of the time, the right structure after a first pass is a mixture rather than a choice.
The bulk of your ongoing practice should be fresh, unseen questions, because unseen performance is the only performance that predicts anything, and because new material keeps exposing weaknesses your first pass missed.
Layered on top, run spaced returns of the concepts you got wrong, using different questions that test the same principles. This is where an adaptive engine and spaced repetition earn their keep, because the machinery for "bring back the concept I failed, in a new form, at increasing intervals" is exactly what they are built to do.
And keep some of your original incorrects for one specific purpose: not to re-attempt, but to audit. Read them again, look at the discriminator you missed, and check that the rule you extracted is one you still hold.
The two variables that shift the balance
How you weight that mixture depends on where you are.
Time to the exam. With months left, favour fresh material and broad coverage, because you have time to expose and repair weaknesses you have not yet discovered. In the final fortnight, shift decisively towards consolidating what you have already identified, plus timed full-length practice, because there is no longer time to open new fronts and there is enormous value in making the known corrections stick.
Your baseline. If your first pass showed a solid baseline with a few clear weak domains, target hard and go narrow: the marginal return is highest in the weak areas. If it showed a weak baseline across the board, you have a coverage and content problem rather than a targeting problem, and broad continued practice with heavy content revision is the honest answer, uncomfortable as that is.
Do not buy a second bank to escape a plateau
One warning. Candidates in a recognition plateau frequently respond by purchasing a second question bank, which does deliver fresh material and does produce a temporary sense of progress.
That is fine if the diagnosis is genuinely "I have exhausted the unseen material in my bank". It is a waste of money if the actual problem is that you are reviewing passively and not converting errors into rules, because you will do exactly the same thing to the new bank, and arrive at the same plateau four weeks later with less time remaining.
Fix the review method first. Then, if you genuinely need more unseen material, get more unseen material.
Where iatroX fits
iatroX is built around the item-versus-concept distinction: the adaptive engine returns the concept you got wrong in a different question rather than serving the same item again, which is the difference between testing your memory and testing your understanding, and spaced repetition schedules those returns at intervals that fight decay rather than letting a week-three correction quietly expire. First-attempt accuracy on unseen questions is tracked distinctly from performance on repeats, so you can see immediately whether your rising score is knowledge or recognition. Try it with free sample questions at iatroX. For diagnosing a score that has stopped moving, see the question bank plateau.
Frequently asked questions
Should I do a second pass through my question bank? Generally no. Re-attempting seen questions measures recognition rather than knowledge, and produces an inflated score that convinces you that you are readier than you are. Prioritise fresh unseen questions, with spaced returns of the concepts you got wrong.
Is it enough to redo only my incorrect questions? It is better than redoing everything, but it still tests whether you remember the answer to that item rather than whether you understood the principle. Return to the concept in a different question instead.
Should I buy a second question bank? Only if you have genuinely exhausted the unseen material in your first one. If your real problem is passive review, a new bank will produce the same plateau a month later with less time left.
How does this change close to the exam? Shift from opening new fronts to consolidating known corrections, and increase timed full-length practice. With months left, favour fresh material and broad coverage; in the final fortnight, favour making the corrections stick.