Safety ratings influence purchasing substantially, and what they test is more specific than a star count suggests.
The testing organisations
Independent and regulatory bodies conducting standardised crash tests.
Which publish protocols openly.
Different programmes use different tests, so ratings are not directly comparable between them.
The test scenarios
Frontal impact, side impact, pole impact and, increasingly, tests for vulnerable road users.
Which represent common serious crash types.
Small overlap frontal tests were added after analysis showed a crash type poorly addressed by existing designs.
What is measured
Forces on instrumented dummies representing occupants of different sizes.
Which are compared against injury thresholds derived from biomechanical research.
Dummy design has expanded to represent a wider range of body types, which was a documented gap.
Active safety
Systems preventing crashes rather than mitigating them.
Which now form a substantial part of ratings.
Automatic emergency braking has strong evidence for reducing rear-end collisions.
Ratings within class
Tests are conducted at defined speeds, which means a small car and a large one rated identically are not equally safe in a collision between them.
Which follows from physics rather than from testing failure.
Mass difference matters, and ratings do not capture it across categories.
Protocol evolution
Requirements are tightened periodically.
Which means a rating from years ago is not comparable to a current one.
The test year should always accompany a rating.
Real-world data
Insurance and casualty data validates whether test performance translates.
Which has generally supported the tests and has identified gaps.
Using ratings
Check the year, the programme, and the individual category scores rather than the overall stars.
Pedestrian protection and child occupant scores are reported separately and matter for different buyers.
Assistance systems and testing
Automated braking, lane keeping and speed assistance are tested in defined scenarios.
Which cannot cover every real-world situation.
Performance in scenarios not tested varies considerably between systems.
Child restraints
Rated separately, including ease of correct installation.
Which matters because incorrect installation is common and substantially reduces protection.
Standardised attachment systems were introduced to address this.
Vulnerable road users
Pedestrian and cyclist protection is now a scored component.
Which has driven changes to bonnet design and to sensor systems.
Vehicle height and front-end design affect this measurably.
Vehicle size trends
Average vehicle mass and height have increased in many markets.
Which has documented consequences for people outside the vehicle.
Research on this has prompted policy discussion in several countries.
Using ratings well
Compare within a size class, check the test year and read the component scores.
Older vehicles
A vehicle rated highly a decade ago would generally not achieve the same rating today.
Which matters when buying used.
Real-world casualty data by vehicle age shows a substantial and consistent gradient.
Retrofit and aftermarket
Modifications can affect crash performance.
Which includes aftermarket bull bars, suspension changes and seat modifications.
Insurance and regulatory implications follow from modifications and are frequently overlooked.
Electric vehicles
Battery placement affects centre of gravity and structural design.
Which has generally produced good crash test performance.
Mass is higher, which has consequences for other road users in collisions.
Maintenance and tyres
Tyre condition affects stopping distance more than most safety features.
Which is entirely within the owner's control and is frequently neglected.
Tread depth and pressure are the two checks that matter most.
Where to look
Testing organisations publish full reports with individual test results, video and scoring breakdowns.
Why ratings improved so quickly
Manufacturers design to the tests, which is exactly what the programmes intend.
Which means protocols must keep advancing to remain discriminating.
The effect on real-world casualties has been substantial and is documented in national statistics.
What buyers should actually check
Test year, programme, component scores and whether the specification tested matches the one being sold, since equipment varies by market and trim level.
A closing observation
Crash testing programmes are among the clearer examples of consumer information changing an industry.
Manufacturers designed to the tests, ratings improved, and real-world casualty figures fell substantially over the same period.
The protocols keep advancing precisely because the industry catches up to them.
Insurance ratings
Separate assessments of repair cost and theft risk that determine insurance groupings.
Which is a different question from crash safety and affects running costs directly.
Both are published and are worth checking together before purchase.
Recalls
Safety recalls are published by manufacturers and by regulators and are searchable by vehicle identification number.
Which is worth checking on any used purchase.
Outstanding recalls on second-hand vehicles are common and are repaired free of charge.
A final note
Ratings compare vehicles tested to the same protocol in the same period.
Any comparison across protocols, across years or across size classes is doing something the rating was not designed to support.
The programmes publish full test reports and video for every vehicle assessed, which is considerably more informative than the star count.