MCAT practice question volume: you are chasing a number, you are hitting a supply ceiling.

The official question pool is a number you can count

Most advice on this treats practice questions as if supply were infinite, as if “just do more” were always available. It is not. The high-quality, exam-representative pool is small enough to put in a table. Here is the whole drillable universe outside the real exam.

Call it under 5,000 distinct, high-quality, non-exam questions in existence. Strong applicants routinely do essentially all of it. The full-lengths and a slice of UWorld are also your calibration tools, so they get used carefully and saved for the right weeks. The supply does not refill. When it is gone, it is gone.

What “more volume” becomes once the bank is empty

Practice questions raise scores through one specific mechanism: retrieval. You pull an answer out of an empty head, and the act of pulling strengthens the memory. That is the testing effect, and it is the entire reason question drilling beats re-reading.

Retrieval only fires when the question is unfamiliar enough that you actually have to retrieve. The moment your eye lands on a stem it has seen before, you are not retrieving anymore. You are confirming. Premeds on the Student Doctor Network forums describe recognizing roughly a third of UWorld questions on a redo. That is the quiet failure mode of chasing volume: the count keeps going up while the share of reps that are genuine retrieval keeps going down.

But redoing questions still works, doesn’t it?

Partly, and it is worth being honest about where it works. Redoing the questions you got wrong is genuinely valuable. You did not know that concept the first time, so even a familiar stem still forces most of the reasoning. Re-reading the explanation on a question you missed is real review. Nobody should stop doing that.

The leak is the questions you got right. Those are the ones recognition flatters. You see the stem, you remember the answer, you mark it correct, and your accuracy stat ticks up while your actual mastery of the concept did not change. Do that across a full reset and you build a comfortable, slightly false picture of where you stand two weeks before test day.

So the fix is not to stop redoing questions. The fix is to change what a redo feels like. If the second pass does not let your eye recognize the stem, a redo becomes a real retrieval rep again, and the supply ceiling stops being a wall.

The volume that actually counts, and where to get it

Split your practice supply into two jobs. The finite official pool, UWorld and the AAMC material, is your calibration: it tells you whether your score is on track, and it works best done once, cleanly, in the right weeks. That supply is not where extra volume comes from.

The volume comes from the content review material you already own. Your Kaplan or Princeton Review chapters, your prerequisite course notes, a recorded lecture: every one of those is a source of practice questions you have not used yet. You do not have to be a prisoner of how many questions AAMC happened to publish. You can generate as many as the content can support.

That is the job Studyly does. Drop in one chapter, one lecture deck, or one PDF and it returns about 200 multiple-choice questions in roughly 60 seconds. The part that matters for the volume problem: every question is auto-rephrased on revisit. When a question comes back around, the stem and the distractors are worded differently, so your eye cannot pattern-match the first three words to an answer you saw three days ago. A redo reads as a new question. That converts the recognition trap, the roughly one-in-three familiar stems, back into genuine retrieval reps.

Quality is the catch with any generated volume. A wrong keyed answer teaches the wrong concept; a giveaway distractor inflates your accuracy. The 81.3 score is the reason generated questions can count as real volume here rather than padding. It is an internal benchmark, not an independent audit, and it does not replace your AAMC full-lengths for score prediction. It does mean the questions you drill during content review are testing the right thing.

A volume plan that respects the supply ceiling

Putting it together, the plan is not a target question count. It is an order of operations.

1. During content review, generate practice from each chapter or lecture as you finish it. This is your high-volume phase, and the supply scales with the material you own, not with what AAMC published.
2. Keep redoing generated questions. Because the wording changes on revisit, those redos stay retrieval reps instead of decaying into recognition.
3. Reserve UWorld for the middle stretch, done once and reviewed hard, focused on the questions you miss.
4. Save the AAMC full-lengths and Section Banks for the final five to six weeks. These are calibration. Do not burn them early chasing a volume number.

The number of questions you do stops being the metric. The metric becomes how many of your reps were real retrievals on content you were weak on. That number you can keep raising right up to test day.