The real-conditions test: sit the exam before the exam

The real-conditions test reproduces the constraint, not just the content. This page sets out exactly what it does: the formats available, where the clock runs, what is monitored during the exam, what the score means and what it does not.

Four formats, from a quarter of an hour to the full exam

The format settles three things at once: how many questions, how much time, and which tests the questions are drawn from. Once an attempt has started, none of this can be changed. That is the whole point of the exercise.

The two short formats cover a single test of your choosing. The half exam mixes the five tests in a fixed proportion, the one we use to model an AD 5 exam: verbal reasoning 35 per cent, EU knowledge 25, digital skills 20, numerical reasoning 10, abstract reasoning 10. The full exam follows the split set out in the competition notice. If you choose a competition, the tests listed in its notice and their share of questions apply instead, including the field-related MCQ test.

Questions are drawn at random from the published bank in your language, every time you start. Two tests in the same format therefore do not contain the same questions.

• Quick: 12 questions in 12 minutes, on the test of your choice.
• Test: 28 questions in 35 minutes, on the test of your choice.
• Half exam: 60 questions in 1 h 15 min, all five tests, weighted.
• Full exam: 110 questions in 2 h 15 min, split as in notice EPSO/AD/427/26: verbal reasoning 20, numerical 10, abstract 10, EU knowledge 30, digital skills 40.

The clock is on the server, not in your browser

At the start, the server writes an end time into the database: start time plus the duration of the format. The countdown on your screen is only a mirror of that value. Moving your computer’s clock forward, suspending the tab or cutting the network does not move that end time.

Any answer sent after the end time is refused and the attempt is marked as expired. You keep the answers already submitted, and the score is computed on the questions you had answered.

Reloading the page loses nothing: the full state is stored in the database, and the page is restored from the record held on the server.

• Only one attempt can be in progress at a time. Starting a test while an attempt is running takes you back into it instead of opening a second one.
• Answers are recorded one by one as the exam progresses. A disconnection does not lose the questions already answered.
• You move forward question by question, with no going back: a submitted answer is final.
• A 5-second countdown precedes the first question, giving fullscreen mode time to settle.

Mandatory fullscreen, and what gets logged

The exam runs in fullscreen. The switch is requested at the very moment you click to start, because browsers only allow it on an explicit user gesture.

Two gestures end the attempt: leaving fullscreen, or moving to another tab or window. In both cases the first occurrence triggers a visible warning and the second invalidates the attempt on the server. In August 2026, switching tabs was brought into line with leaving fullscreen: until then, staying in fullscreen while switching to another window had been the simplest way around the whole system.

Other gestures are blocked and logged without invalidating anything. They are deterrents that add friction, not proof of cheating.

• Copy, paste and cut: blocked, counted and timestamped in the attempt log.
• Right-click: blocked and logged.
• Refresh shortcuts and shortcuts that open or close tabs: intercepted as far as the browser allows.
• Closing the page: the browser shows its own confirmation prompt.
• The log is capped at 200 events per attempt. Beyond that, events are ignored and the counters are frozen.
• What we do not block, and do not claim to block: browser developer tools, a sheet of paper, a second device. A rigged test only fools the person sitting it.

What the result says, and what it does not

At the end, each test gets its success rate as a percentage of correct answers. For the two long formats, an overall score out of 100 applies the weighting described above. For the two short formats, the overall score is simply the success rate of the chosen test.

We show no pass score, and that is deliberate: pass scores are set for each competition in the official notice, and are often supplemented by a ranking based on the number of posts. A score here does not tell you that you would have passed; it tells you where you stand, test by test, under exam conditions.

An invalidated attempt produces no score, and neither does an abandoned one. An expired attempt produces a partial score, computed on the questions answered.

• The time actually spent is shown next to the score.
• Questions left unanswered are listed so you can revisit them one by one, with their explanation.
• Your past attempts stay available: the last 20 are listed, most recent first.
• The test is open to Pro accounts, with one test every 90 days, and to Premium accounts without limit. Since August 2026 the check is made by the server when the attempt is created, and no longer only when the page is displayed.
• The exercise is designed for a computer. On a mobile phone, we offer to email you a link so you can pick it up on a suitable screen.