Hazard perception
Why memorising practice clips fails, and how much hazard perception practice is enough
Last reviewed against gov.uk on
There is a way to practise hazard perception for weeks, know every clip by heart, walk in confident and score in the thirties. It happens to people who worked hard, and it happens for one structural reason: you cannot revise the answers to a test whose questions you have never seen. The clips waiting for you are not in any revision set. What you can carry into the exam room is a skill, and the practice that builds it looks different from the practice that builds recall.
No revision set contains the clips you will sit
Start with the claim that keeps circulating in learner forums — that some practice clips are the same as the ones in the real test. It is not a small exaggeration. It is the opposite of how the material is produced.
The agency licenses revision content to publishers through a formal route, and the terms of that route are published. Both halves of the theory test are covered, and both carry the same caveat.
They’re not the exact clips that are used in the real test, but they’re very similar.
That is the licensing page for commercial revision products, which is the closest a practice clip can get to the live set. Clips that reach you any other way are further from the live set, not closer. The free practice test on GOV.UK says the same thing in plainer words.
These clips are not used in the real test, but they show you how the test works.
So there is no version of this where memorising pays off. Not with paid material, not with free material, not with ours. Every clip you have ever practised on is a rehearsal of the format. None of them is a preview of the paper.
The multiple-choice half works the same way, which is worth knowing if you have been told the questions leak: the licensed revision questions are “not the exact questions that are used in the real test, but they’re very similar”. Similar topics, similar phrasing, different items.
Why memorising fails even on the clips you memorised
Suppose you ignore all that and learn a practice set cold — you can name every scene, you know the cyclist one and the bus one and the one with the tractor. On those very clips, your score goes up. That improvement is real, and it is almost entirely worthless, because of what the score is measuring.
To get a high score, click the mouse as soon as you see the hazard starting to develop.
Points come from when you click relative to the hazard beginning to develop — not from whether you were right about what was going to happen. On a clip you have memorised, you are not detecting anything. You are recalling a timestamp. The mental operation is closer to remembering a phone number than to reading a road, and it produces a high score by a route that does not exist in the exam.
Then there is the failure mode that this produces on test day. A candidate who has trained on recall arrives with a set of expectations — a mental list of scenes — and spends the first few clips searching the screen for a match. That search is slower than simply watching the road, and slow is exactly what the scoring punishes. Worse, when nothing matches, the temptation is to fall back on a fixed routine of clicks per clip, which is the one behaviour that scores nothing at all.
You do not lose points if you click and get it wrong. However, you will not score anything if you click continuously or in a pattern.
Our how many clicks guide covers what that detection actually catches and why a fixed routine trips it.
The proof that the skill is the transferable part
You do not have to take this on argument, because the agency ran the experiment in public. When the filmed clips were replaced with computer-generated ones in January 2015, every candidate in the country had been practising on footage that no longer resembled the test they were about to sit. The official advice at the time was not to start again.
Any products that you’ve been using to help you practise the hazard perception part of the test are still relevant. You do not need to have practised with CGI clips in order to take and pass the test.
Read that carefully and it settles the whole question. Practice built on entirely different footage — different medium, different vehicles, different roads — still prepared you for the test, because what it had trained was the judgement, not the images. Familiarity with specific clips was worth so little that swapping the entire library out did not invalidate anyone’s revision.
The library has moved on again since. Weather and low-light scenarios were added from November 2018 — fog, rain, snow and ice, wind, and driving at night and in low light such as dawn and dusk — and the same principle held.
The way the hazard perception part of the theory test works and is scored is not changing. The pass mark is also not changing.
What your practice score is actually measuring
Every hazard perception clip carries a scoring window: a stretch of time inside which a click scores, decreasing in value the later it lands. That window is data attached to the clip, not something you can see. Licensed revision clips ship with it.
The clips include scoring window timeframes.
This is the part that makes practice scores harder to read than people assume. A clip that came through the licensing route arrives with its window already set. A clip that came from somewhere else has a window chosen by whoever built it — an estimate, made in good faith or otherwise, of where a real examiner’s window would sit. You have no way to tell the two apart from the outside, and the score you see is generated by whichever it is.
That does not make practice pointless. It makes a practice score a readiness signal rather than a prediction.
| What you see in practice | What it reliably tells you | What it does not tell you |
|---|---|---|
| Consistently high scores across clips you have never seen | You are detecting cues early and your clicking has the right shape | Your exact mark on the day — a different window shifts every number |
| High scores only on clips you have run several times | You have learnt those clips | Almost nothing about the test, which contains none of them |
| Scores hovering just above the pass mark | You are clicking late — mid-window rather than at the cue | That you have a margin, because the margin is inside the error |
| Occasional zeroes on clips you clearly read correctly | Your clicking shape is the problem, not your observation | Which clip it will happen on next time |
The arithmetic underneath is unforgiving in a quiet way. The car test runs 14 clips and 15 scored hazards for a maximum of 75 points, and you need 44 — an average of just under 3 out of 5 on every hazard. Clearing that by two or three points on unverifiable clips is not a pass in waiting. It is a coin toss with the full test riding on it, because failing this half means booking and sitting both halves again.
You must book and take the full test again, even if you passed one part this time.
So how much practice is enough?
There is no published number of clips or hours, and anyone quoting you one has made it up. What there is instead is a standard you can actually check yourself against, and it has three parts.
1. Breadth, not repetition. Practise across the conditions the live library contains, and stop re-running clips you have already seen. The conditions are a matter of record: fog, rain, snow and ice, wind, night, and low light such as dawn and dusk, on top of the ordinary urban, rural and dual-carriageway scenes. A learner who has only practised in bright daylight has a genuine gap; a learner who has run the same twenty daylight clips eight times has an imaginary strength.
2. Consistency on first viewing. The only meaningful practice score is the one you get on a clip you have never seen. Keep a mental split between fresh clips and repeats, and judge yourself exclusively on the fresh ones. When your first-viewing scores stop swinging wildly, the skill has stabilised — that is the signal to look for, more than any total.
3. Comfortable margin, not a bare pass. Because you cannot verify anyone’s scoring window, build in headroom. If your fresh-clip average is landing comfortably above the pass mark rather than brushing it, the underlying behaviour — clicking at the cue instead of at the consequence — is probably right, and it will survive a window that sits slightly differently from the one you practised on.
| Builds the skill | Builds recall (and flatters your score) |
|---|---|
| New clips you have not seen before | The same clips run repeatedly until the score rises |
| Naming the cue out loud — “indicator”, “head turn”, “brake lights” | Naming the clip — “this is the tractor one” |
| Varied weather, light and road types | Whatever is at the top of the list, in the same order |
| Reviewing a low score by asking which frame you missed | Reviewing a low score by trying to remember the timing |
| Watching the road as a passenger and calling hazards aloud | More clips, faster, with less attention each |
The two things that add more than volume
Understanding the format properly. When the CGI clips were trialled, the finding was not that people needed more practice. It was narrower and much more useful.
The trial was successful. It found that candidates who watched the new introduction clip about how the test works were more likely to spot the developing hazards.
Watching a short explanation of the mechanic measurably improved detection. If you have not yet nailed down what separates a developing hazard from a static one, an hour on when to click and the hazard perception overview will do more for your score than another fifty clips.
Real road experience. Hazard perception is not a video game skill that happens to be tested on video. It is the thing the test exists to sample, and it improves off-screen.
A 2008 study published by the Department for Transport Cohort II: A study of learner and new drivers found that hazard perception is a trainable skill and that improving it can reduce collision liability.
The same document notes that young and novice drivers tend to lack the skill because of limited experience. Lessons and supervised practice are hazard perception training, whether or not anyone calls them that — and the payoff outlasts the exam.
Since its introduction, the hazard perception test has been shown to reduce non-low-speed accidents by an estimated 11% for novice drivers in their first year of driving.
That figure is the reason the test is scored the way it is. It is not measuring whether you can memorise footage, because memorising footage does not stop anyone crashing.
A practice routine that transfers
- Learn the mechanic first, once. What opens the scoring window, what a pattern response costs you, what a cross at the end of a clip means. This is a single session, not an ongoing activity.
- Run the three free GOV.UK clips deliberately. They are at gov.uk/take-practice-theory-test. Click early on one pass and late on the next, and watch what the timing does. Five minutes here calibrates more than an hour of passive watching.
- Then build volume on clips you have not seen. Keep the run fresh; when you run out of unseen clips, switch to naming cues out loud on the ones you repeat, so you are training detection rather than recall.
- Say the cue, then click. Out loud, in practice: “indicator” — click. It forces the click to be anchored to something observable, which is the habit that survives into an unfamiliar clip.
- Cover the awkward conditions on purpose. Night, rain, spray, low sun. These are in the live library and they are where untrained eyes are slowest.
- Do the road version. As a passenger or on a lesson, call developing hazards as they appear. No clip library reproduces the variety of a real junction.
- Judge readiness on fresh clips only, and give yourself margin above the pass mark before you book.
When you have practised a lot and the score is still low
If fresh-clip scores are stuck, the cause is usually one of three things, and they need different fixes.
- You are clicking at the consequence, not the cue. The commonest by far, and it looks like good observation because you did see it. Fix by rewatching a clip and identifying the earliest frame where a driver’s hands or feet would have moved.
- Your clicking shape is zeroing clips. You read the clip fine but you clicked in a rhythm or a burst. Fix by cutting to one deliberate click plus at most one confirmation.
- You are clicking on the set-up. Parked cars, a bus stop, a blind bend — potential hazards that have not begun to develop. Clicks before the window opens score nothing and, repeated across fourteen clips, start to look like a pattern.
None of those is fixed by more clips. All three are fixed by changing what a click is anchored to.
The questions people actually ask
“Are any practice clips the same as the real test clips?”
No. Even the clips licensed for commercial revision products are not the ones used in the live test, and clips outside that route are further away still. Anyone telling you their set contains real test clips is either mistaken or selling something.
“I memorised a whole practice set and still failed. What went wrong?”
Nothing you did was recorded in the exam. You trained recall of specific footage, and the test presented footage you had never seen and scored how early you reacted to it. It is a common and demoralising outcome, and it is a strategy problem rather than an ability problem.
“How many clips should I practise on before booking?”
There is no official figure. The better test is whether your scores on clips you have never seen are consistent and comfortably above the pass mark, rather than how many clips you have got through.
“My practice app gives me 60 out of 75 — will I get that in the test?”
Treat it as a signal that you are reading clips well, not as a forecast. The score depends on a scoring window you cannot inspect, so the number moves even when your driving judgement does not.
“Does watching dashcam videos help?”
Yes, as skill training — it is one more source of unfamiliar road scenes to call hazards on. It will not teach you the click mechanic or the scoring window, so use it alongside proper clips rather than instead of them.
“Should I practise if I have already passed the multiple-choice part?”
You will be sitting both halves again regardless, because a failed part means the whole test is rebooked. Put the time into hazard perception and keep the questions ticking over — see what happens next.
Where to practise
Three practice clips are free at gov.uk/take-practice-theory-test, which is where to learn the mechanic. For volume on clips you have not seen, our clip gallery shows the range of scenarios in the app — rural lanes, town high streets, motorways, plus snow, night, rain and low sun — and a timed mock test keeps the multiple-choice half sharp at the same time. The app is free to get started, with free DVSA questions and hazard perception clips to get you going.
Practise on scenes you have never seen, click at the cue, and let the number take care of itself.
Sources
- GOV.UK — Make and sell driving theory test revision products Licensed revision questions and hazard perception clips are not the ones used in the live test
- GOV.UK — Take a practice theory test
- GOV.UK — Theory test: hazard perception test
- GOV.UK — Theory test: pass mark and test result
- GOV.UK — Hazard perception clips get a modern makeover (published 18 December 2014; the CGI clips replaced the filmed ones on 12 January 2015)
- GOV.UK — Hazard perception test now features more weather conditions
- GOV.UK — 20 years of the theory test making roads safer
- Department for Transport — Introducing a minimum learning period for learner drivers