How many practice questions is enough for the FACEM Written?
Every registrar prepping for the FACEM Written eventually asks some version of the same question: how many practice questions do I need to do before I sit? Two thousand? Five? The whole bank, twice? I asked it too. It is the wrong question, and chasing a number is one of the easier ways to feel productive while learning almost nothing.
The count of questions you have answered is a vanity metric. It measures hours logged, not knowledge gained. You can grind through four thousand items, feel exhausted and virtuous, and walk into the exam carrying the same blind spots you started with. I have watched strong candidates do exactly that.
Coverage beats volume
The reason a large bank matters is not that you will answer all of it. It is that a large, well built bank covers the blueprint properly: every body system, the common presentations and the uncommon-but-lethal ones, the toxicology you never see on a normal shift. A bank of over 4,000, growing weekly, gives you the surface area to find your weak spots. That is its job. It is a map of the territory, not a step counter.
So the useful question is not “how many have I done” but “how much of the blueprint have I actually met, and how did I go on it”. A candidate who has worked carefully through 1,500 questions spread evenly across every topic is better prepared than one who has smashed 3,000 in the three systems they already enjoy.
The signal that actually predicts a pass
If I could give you one number to watch, it would be first-attempt accuracy, broken down by topic. Not your overall percentage. Not your total answered. Your first-attempt accuracy per subtopic.
First attempts are honest because you cannot fake them. The second time you see a question you half-remember the answer, and your score drifts up for reasons that have nothing to do with what you will know in the exam hall. First attempts, tracked separately, tell you the truth about where you actually stand. When your paediatric ECG accuracy is sitting at 55 per cent and your sepsis accuracy is at 85, you know precisely where the next fortnight should go. That is worth more than any headline count.
This is also why timed mode matters as the exam approaches. Feedback withheld to the end, no explanation to lean on, forces a genuine first attempt under pressure. It is uncomfortable. It is meant to be.
Your incorrect pile is the syllabus
Say you work through a couple of thousand questions and get roughly 300 wrong. Those 300 are the most valuable thing you own. They are a personalised syllabus, written by your own gaps, handed to you for free.
Most people do the opposite of what they should. They answer a question, get it wrong, read the explanation for ten seconds, nod, and move to the next fresh item because the counter goes up when they do. The learning is in going back. Build a session from nothing but your incorrects. Work each one back to the underlying principle, not the surface fact. Why is that the key. Why is each distractor wrong. What would change the answer. Then leave it a week and see the same question cold. If you still get it, it has moved from recognition to knowledge.
Two hundred questions genuinely worked back through will do more for your result than two thousand answered once and forgotten.
Depth over count
The five-part explanation on a good question is where the actual teaching lives: why the key is right, why every distractor is wrong, the one-line bottom line, one step further for depth, and the source. Reading all five on a question you got wrong teaches you more than the next twenty you rush through. Depth is slower. It is also the only thing that survives contact with the exam.
So, how many is enough?
Enough is when your first-attempt accuracy is where it needs to be across every topic, and your incorrect pile has shrunk because you actually closed the gaps rather than papering over them. For some people that is 1,500 questions. For others it is 3,000. The count is an output, not a target, and it is different for everyone.
Stop counting how many you have done. Start watching how well you do them the first time, and go back to the ones you got wrong. That is the whole game.
If you want to see what a properly built, structured-explanation question looks like, a free sample set is here: try the samples.