Three kinds of hazard, and only one of them scores
Almost every mark lost on this test is lost by clicking on the wrong sort of thing, and the distinction is never explained properly because it sounds like pedantry until you see the score.
A static hazard is a permanent feature of the road: a bend, a narrowing, a parked row, a junction. It is there in the first frame and it will be there in the last. Clicking on it scores nothing at any point in the clip.
A potential hazard is something that could become your problem and currently is not. A cyclist riding along the kerb. Somebody standing at a crossing. A car waiting properly at a give way line. You should be watching all of these and easing off for some of them, and clicking on them still scores nothing.
A developing hazard is the one the test pays for, and gov.uk defines it in a single sentence: something that would cause you to take action, like changing speed or direction. The cyclist pulling out. The pedestrian stepping on. The waiting car nosing across your lane. Nothing about the first two frames of any of those makes you brake. The moment it changes is the moment worth five marks.
Why being early is not the same as being good
The instinct on a practice run is to click the moment you notice anything, on the reasoning that early must be safer than late. That habit alone can cost you the test.
A click before the hazard has started developing is outside the window and scores zero. So a driver with genuinely excellent observation, who spots the parked cars and the school sign and clicks on both, can come away with nothing from a clip they read perfectly. What is being measured is not vigilance, it is the specific moment a situation turned.
This drill shows you the window on a bar after every clip, with your clicks marked on it. Seeing three clicks sitting to the left of the green band explains more in one glance than any amount of being told.
Clicking continuously or in a pattern scores nothing at all
The most useful line on the whole subject, and it is a direct quote: you will not score anything if you click continuously or in a pattern.
That is per clip, and it is absolute. Spreading clicks across the clip in the hope of catching the window does not give you a worse score, it gives you no score, even if one of them landed perfectly. A steady rhythm does the same thing. The rule exists because otherwise a metronome would beat an attentive driver, and it catches people who have never been told it is there.
This page enforces it and says so when it fires. The exact algorithm DVSA uses is not published, so the one here is ours: more than eight clicks in a clip, or five or more whose gaps barely vary. What transfers is not the threshold, it is that the strategy does not work.
The pass mark is softer than the other half, and almost nobody knows
A car theory test is two halves and they are not equally demanding. The multiple choice wants 43 out of 50, which is 86 per cent. Hazard perception wants 44 out of 75, which is under 59.
You still have to pass both in the same sitting, so the softer mark is no comfort if you fail it. But it is worth knowing which half you are closer to, because the revision is completely different: one is reading and the other is a reaction you cannot look up.
These are drawings, and here is exactly what that costs you
Nothing on this page is footage. The scenes are vector shapes drawn from a script, with a real perspective camera behind them so distance and closing speed behave properly.
What that does not give you is the visual clutter of a real street, and clutter is part of the difficulty. A drawn scene has nothing in it that we did not put there, so spotting the moving thing is easier than it will be on the day.
What it does give you is teachable: the three way distinction above, the shape of the scoring window, and the pattern rule. It also gives you something no clip can, which is the window drawn out afterwards so you can see precisely how early or late you were.
A clip only works once, and nobody in this market admits it
This is the honest problem with all hazard perception practice, free or paid. Once you know when the hazard arrives, you cannot un-know it. A second run measures your memory of the clip.
A question bank degrades gently as it repeats: you have seen the question before but you still have to know the answer. A clip does not degrade, it dies. So this page keeps count of which scenes you have been shown, gives you unseen ones first, and when they are gone it tells you plainly that the score has stopped meaning anything. Nothing is uploaded to do that, it is a list of ten names in your own browser.
Which leads to the only honest advice about the real thing: watch the official example clip on gov.uk before you book, so the format is not new to you, and treat any practice clip you own as one attempt rather than a revision loop.
Common questions
Are these the real hazard perception clips?
No, and they are not even video. Every scene here is drawn in your browser from a script. That is a deliberate choice rather than a shortcut: DVSA does not publish where the score boundaries fall inside its own clips, so anybody selling you a reproduction of them is guessing. Scoring a scene we wrote ourselves involves no guess, and both ends of the window are shown to you after every clip.
What is the real test?
For cars and motorcycles, 14 video clips containing 15 developing hazards, up to 5 points each, so 75 available and a pass mark of 44. For lorries, buses and coaches it is 19 clips and 20 developing hazards, 100 available and a pass mark of 67. One of the clips in each case has two developing hazards in it rather than one.
Why does clicking early score nothing?
Because the test scores when you spotted a hazard developing, not when you spotted something that might one day become one. gov.uk defines a developing hazard as "something that would cause you to take action, like changing speed or direction". A parked car does not make you do that. The door opening does. Click on the parked car and you have clicked before the window opened, which scores zero however alert it felt.
Can I just click a lot to be safe?
No, and this rule costs people whole tests. gov.uk states it plainly: "you will not score anything if you click continuously or in a pattern." That is per clip, so a spread of clicks that happens to include a perfectly timed one still scores nothing. This drill enforces the same thing and tells you when it has.
Why does it count down how many scenes I have left?
Because a hazard perception clip only works once. Once you know a child steps out from behind the third parked car at five seconds, you can score full marks on that clip forever without watching it, and you have practised nothing. This is true of every hazard perception product including the paid ones, and none of them say so. When you have seen all ten, the page tells you the mark has stopped measuring your observation.
How is this scored?
The same five points per hazard the real test uses, on a window that opens the moment the hazard starts to develop and closes when it is too late to matter, cut into five bands with the earliest worth most. The pass mark is the real one scaled to the length of the drill: 44 out of 75 becomes 30 out of 50 across ten scenes.
Can I go back and change a clip?
No, and neither can you in the real test: "You only get one attempt at each clip. You cannot review or change your responses." That is why this shows you the window and the teaching straight after each clip rather than at the end, which is the one place practice can be more useful than the exam.