Safety ratings are quoted as a simple number and rest on a specific set of tests that are revised periodically.
The tests
Frontal, side and pole impacts, plus pedestrian protection and assistance systems.
Which are weighted differently by different rating organisations.
Comparing across years
A five-star rating from a decade ago used easier criteria.
Which makes cross-year comparison misleading.
Weight and compatibility
Tests are conducted against fixed barriers or similar vehicles.
Which does not capture what happens between vehicles of very different mass.
Occupant protection
Measured through instrumented dummies recording forces.
Which represents an average adult occupant.
Smaller occupants and different seating positions are less well represented.
Child protection
Separate assessment with child restraints fitted.
Which is scored independently and frequently overlooked by buyers.
Assistance systems in the score
Automatic emergency braking and lane support now contribute.
Which has shifted what a top rating requires.
What ratings cannot capture
Real-world crash variety, maintenance state and driver behaviour.
Which is why a rating is a comparison rather than a guarantee.
Using them properly
Compare vehicles tested under the same protocol in the same period.
Different rating bodies
Regional programmes with different protocols and scoring.
Which means a rating from one region does not transfer directly to another.
The same model can be built to different specifications for different markets.
Protocol revisions
Criteria tightened every few years.
Which is why older ratings look generous by current standards.
Real-world crash data
Insurance and government casualty statistics.
Which complement the test results and reflect actual use.
The two broadly agree, which is reassuring for the test protocols.
Beyond the rating
Visibility, seating position, headlight performance and ergonomics.
Which affect crash avoidance and are only partly captured.
Using ratings sensibly
A high rating is a good filter and is not the whole assessment.
How the tests are conducted
Standardised impacts at defined speeds against defined barriers.
Which produces repeatable, comparable results.
Repeatability is the reason the tests do not resemble typical real crashes.
Design to the test
Manufacturers optimising for the specific test protocol.
Which is a known criticism and part of why protocols change.
Rating bodies add new tests partly to counter it.
Small overlap and oblique impacts
Tests added after real-world data showed a gap.
Which prompted significant structural redesign across the industry.
Headlight assessment
Included by some rating bodies after finding large real-world differences.
Which affects night-time crash rates measurably.
Using ratings when buying used
Check the rating and the year of the protocol.
Which together tell you what the score actually represents.
Ratings for older vehicles
Many vehicles on the road predate current protocols entirely.
Which is relevant when buying at the cheaper end of the market.
Structural safety has improved substantially over two decades.
Motorcycle equivalents
No comparable structural rating exists.
Which shifts the emphasis onto rider equipment and anti-lock braking.
Commercial vehicles
Separate assessment programmes with different priorities.
Which include direct vision standards in some cities.
What to prioritise as a buyer
Automatic emergency braking, good headlights, a recent rating and a vehicle that fits you properly.
Which together matter more than a single star count.
Where to find the data
Rating organisations publish full test results rather than only the star count.
Which includes individual scores by category and is free to access.
The detail is more useful than the headline, particularly for child occupant protection.
The summary
A rating compares vehicles against a protocol from a particular year, protocols tighten over time, and the score is a useful filter rather than a complete assessment of how safe a vehicle is in real use.
A general note on sources
Figures in this area come from manufacturer publications, regulator test programmes, insurance claims data and independent consumer testing, and those four sources do not always agree.
Where they conflict, the independent testing and the real-world claims data are usually the more reliable guide, because manufacturer figures are produced under conditions chosen by the manufacturer.
Anything specific to your own vehicle should be checked against its handbook and against a qualified technician familiar with the model, since specifications differ between markets and between production years in ways that general articles cannot capture.
One last practical point
Almost everything above becomes easier if you keep a simple record for your own vehicle: what was done, when, at what mileage and by whom.
It takes a folder and about five minutes a year, and it improves resale value, makes warranty claims straightforward and turns vague worries about condition into questions you can actually answer.
Owners who do this rarely get caught out, and owners who do not almost always wish they had started earlier.
Where to go for model-specific detail
Owner forums and marque specialists know more about any individual model than any general article can, including which components fail and roughly what the repair costs.
An hour reading before a purchase or a repair decision is consistently one of the better uses of time available to a vehicle owner.