How to sit a long paper without falling apart in the second hour

Most people who do worse than they should in a long exam don't fail because they didn't know enough. They fail in the middle. The first hour goes well. Somewhere in the second, concentration thins out, a hard stretch of questions eats ten minutes you didn't plan to spend, panic appears, and the next thirty questions are answered by a person who is not quite the one who prepared.

That's a skill problem, not a knowledge problem, and like any skill it improves with practice. This article covers how to practise it. It's about the exam as an event: pacing, deciding what to skip, whether to guess, how to protect your attention, and how to learn from a mock test properly.

A note before we start. Exam details differ by year, so everything about the marking, the number of questions, the time and the sections is something you should confirm in the official bulletin for your cycle. We'll use the commonly reported scheme (four marks for a correct answer, one mark deducted for a wrong answer, nothing for a blank) to explain the reasoning, but do the same arithmetic with whatever the bulletin actually says.

Know the clock before you start

The first piece of pacing is arithmetic. Divide the total time by the number of questions. If a paper gives you roughly three and a half hours for a couple of hundred questions, that's a little over a minute per question on average. Whatever your exam's number is, write it down.

Now the important part: an average is not a budget for every question. Some questions take twenty seconds, because you either know the answer or you don't. Some take two minutes, because they are long vignettes with a lab table. If you spend your average on each, you'll run out of time on the long ones and have time left over on the short ones.

Think of the clock as a savings account. Fast questions deposit time. Hard questions withdraw it. Your job is to make sure that, at checkpoints, the balance is where it should be.

Set checkpoints, not just a finish line

Pick two or three checkpoints and write them on your rough sheet at the start. If you're given sections, use the sections. If not, use quarters. At each checkpoint you want to have reached a certain question number. If you're behind, you speed up. If you're ahead, you have a little room to spend on the hard ones.

Checkpoints solve the biggest pacing problem, which is that time disappears gradually and you don't notice until it's gone.

The three-pass method

A very common approach to a long MCQ paper, and one that tends to work, is to go through it in passes.

Pass one: take the sure marks. Go through quickly. Answer anything you know within about a minute. For anything that looks long, confusing or unfamiliar, flag it and move on without agonising. Aim to finish this pass with time to spare.

Pass two: the maybes. Return to the flagged questions. Now you have a sense of the whole paper, and you've banked easy marks, so you can think with less pressure. Spend your time here on the questions where you can narrow to two options and need a bit of thought.

Pass three: the rest. For whatever remains, make your best call. Don't leave a question blank until you've considered whether guessing makes sense (more on that below).

This method does two helpful things. It protects the easy marks from being squeezed out by one tough question, and it takes the sting out of not knowing something, because "flag and move on" is a plan and not a failure.

What to flag

Flag a question if you can't make progress within about a minute, if it's a long vignette that will take work, or if you've narrowed to two and aren't sure. Don't flag questions you know. And don't flag everything: if half your paper is flagged, you haven't been decisive in pass one.

Should you guess?

The honest answer is "it depends on the marking", so let's do the arithmetic once, with the commonly reported scheme.

Suppose a correct answer earns 4 marks and a wrong answer costs 1. With four options, a blind guess has a one-in-four chance of being right. The expected value is 0.25 × 4 minus 0.75 × 1, which is +0.25 marks. A tiny positive number. If you can eliminate one option, your chance is one in three, and the expected value rises to (⅓ × 4) − (⅔ × 1), about +0.67. If you can eliminate two options, it's 50:50, and the expected value is +1.5.

So under that scheme, narrowing down is worth a lot, and even a blind guess isn't, on average, a loss. That's an argument for answering most things, especially after you've eliminated an option or two.

There are two caveats, and both matter.

First, check the real scheme. If the bulletin for your year says something different, redo the arithmetic. The conclusion depends on the numbers.

Second, expectation is not the whole story. A strategy with a small positive average still has variance. A string of unlucky guesses will cost marks, and if you're a person who gets rattled by negative marks, that anxiety has a cost of its own. A sensible middle path is: guess when you've eliminated at least one option and have a reason to prefer one of the rest; leave a question blank only when you have genuinely no information and you know that a loss will shake you.

Also remember that "I know it, I'm just not sure" is not the same as "I have no idea". Most uncertainty is partial knowledge, and partial knowledge is worth acting on.

Reading the question the way the examiner wrote it

Many lost marks in a long paper are reading errors, not knowledge errors. A few habits help.

Read the last line first. If the question ends with "which of the following is NOT", you read the vignette differently. Knowing what is being asked changes what you look for.

Underline the key words. If you have a rough sheet, write the diagnosis or the clue in a word or two. "ACE-i, cough" is enough.

Watch for the next step versus the diagnosis. A vignette might describe an emergency and ask for the initial management, not the final treatment. The answer to "what is the most important first step?" is often not the most sophisticated option. It's the one that stabilises the patient.

Watch for absolutes. Options with "always" or "never" are rarely right in medicine, though sometimes they are. Treat them with suspicion, not as a rule.

Don't invent facts. Answer using what is in the question. If the stem doesn't mention a complication, don't assume one exists.

Stamina is a skill you can train

Three and a half hours of concentrated reading is a different thing from two hours of casual study. Attention thins. Your eyes tire. If you've never sat for that long, the second half of the paper is the first time you meet your own limits, which is a bad moment to do it.

The fix is simple and not exciting: practise at the length of the real thing. Do at least a few full-length timed sittings before the exam. Do them on a computer, at roughly the same time of day as the real paper, in a quiet room, with no phone. The first one will probably feel awful. The second will be a little better. By the fourth, the length itself stops being scary.

A few more things that help:

• Sleep matters more than one more night of revision. For the last two nights, protect your sleep even if it feels like wasted time. A tired brain makes reading errors.
• Eat something sensible beforehand and carry water if you're allowed to. A sugar crash in the second hour is real.
• Plan one micro-break. If the interface or rules allow, a moment to close your eyes, roll your shoulders and breathe can reset your focus. Even ten seconds helps.
• Do the logistics early. Know where the centre is, how long it takes to get there, what documents are needed. Every unanswered logistical question is a source of background worry.

What to do when you hit a bad patch

It will happen. You'll meet five questions in a row that you can't answer, and a voice in your head will say that you've prepared badly and the whole paper is going to be like this.

Some things to remember. The paper is not uniformly hard for you: it's a random sampling of everything, and bad stretches happen by chance. Other candidates are meeting the same difficulty. And your job is only to answer the next question as well as you can.

A tiny practical move: when you notice panic, stop for one breath, then write the number of the next question on your rough sheet, as though starting afresh. It sounds silly and it works, because it separates the next question from the last five.

Mock tests: using them properly

Mocks are how you practise all of the above, but the way most people use them wastes most of their value. They take a mock, look at the score, feel something, and move on. The score is the least interesting output.

Here is how to review a mock so that it changes your next one.

1. Look at pacing first. Where did you reach your checkpoints early or late? Which stretch of questions consumed a lot of time? Did the second half get worse?

2. Sort your misses by reason. Use the same categories as for any question: a gap in knowledge, a misread, narrowed to two and picked wrong, changed a right answer, ran out of time. Tally them. The tally is more useful than the score.

3. Check your guesses. Of the questions you guessed, how many did you get right? If you're consistently below chance, you're guessing carelessly. If you're above, your partial knowledge is better than you thought.

4. Check the questions you flagged. Did flagging help? Were the flagged questions mostly the ones you got wrong, or did you flag questions you'd have answered fine?

5. Pick two fixes. Not ten. Two things you'll do differently next time, one about content and one about technique.

6. Wait before you retry. Go back to the missed topics after a few days, not straight away.

Be careful with small, noisy results

One mock is one morning. A score that jumps ten points or drops ten points between two mocks might be real progress, or might be the difference between a paper that suited you and one that didn't. Look at the trend over several mocks, and look at subject-level patterns only when you have enough questions to say anything. NextPrep shows the number of attempts beside every percentage for exactly this reason, and it doesn't predict ranks. Nobody can predict a rank from a handful of practice papers, and anyone who claims to is guessing.

Also keep in mind that mock papers are written by different people, with different difficulty. A harder mock and a lower score is not a step backward. Compare like with like.

The last week

Don't start new topics. Revise what you've already seen, especially the repeated mistakes in your log and the cards that keep coming back. Do one or two lighter practice sets, not marathon ones. Reread your checkpoint plan. Confirm logistics. Sleep.

On the day, do your first pass quickly, flag without guilt, and trust the process you've rehearsed. You are not trying to be brilliant for three and a half hours. You're trying to be steady.

A short checklist

• Confirm the real pattern, marking and timing in the official bulletin.
• Work out minutes per question and write your checkpoints before you begin.
• Use the three passes: sure marks, maybes, the rest.
• Do the guessing arithmetic with the real scheme, and guess with partial information.
• Read the last line of the question first.
• Practise at full length, on a computer, at the right time of day.
• Review each mock for pacing and reasons, not just the score.
• Protect your sleep in the final nights.

NextPrep has grand mocks, stage tests and subject tests built from its sample questions, with a countdown, a question palette, and a mark-for-review button for the flagging habit. They are practice papers written for NextPrep, not official papers, and the questions are still pending expert review. Treat them as a way to rehearse the skill of sitting a long paper. The knowledge is only half of what you're being tested on.