A five-star safety rating sounds absolute. It reflects performance on a defined set of tests, which differ between testing organisations and change over time.
What gets tested
Frontal impact, generally both full-width and offset, since the two load the structure differently.
Side impact, both from a moving barrier and against a pole, which concentrates force on a small area.
Rear impact assessment for whiplash protection.
Pedestrian protection, assessing the front of the vehicle against head and leg impact.
And increasingly, assessment of active safety systems — automatic braking, lane keeping, driver monitoring.
Why ratings differ between regions
Testing organisations in different markets use different protocols, different speeds and different weightings.
Which means the same vehicle can hold different ratings in different regions, and comparing across them is not meaningful.
Protocols also tighten over time, so a five-star rating from several years ago represents a lower standard than a current one.
The year of the test is stated and is routinely ignored in marketing.
The mass problem
The most important limitation and the least understood.
Crash tests generally assess a vehicle against a barrier or against a standardised object.
They do not assess it against another vehicle of different mass.
In a collision between vehicles of different mass, the occupants of the lighter vehicle experience greater forces, regardless of either vehicle's rating.
Which means a five-star small car and a five-star large vehicle are both rated against their own class, and the ratings do not describe what happens when they meet.
This is a known limitation and testing organisations state it, and it does not survive into consumer understanding.
The weight escalation
Vehicles have become substantially heavier over decades, driven by size, equipment and, in electric vehicles, batteries.
Which improves occupant protection in the heavy vehicle and worsens outcomes for everyone else, including pedestrians.
Research examining collision outcomes finds this effect clearly, and it has begun to influence regulation in some jurisdictions.
Front-end height is a related factor, with taller fronts producing worse pedestrian outcomes because impact occurs higher on the body.
Active safety
The area where the largest gains have come recently.
Automatic emergency braking has been shown to reduce rear-end collisions substantially in real-world data, which is stronger evidence than test performance.
Systems vary considerably in capability — some detect only vehicles, others pedestrians and cyclists, and performance in darkness varies enormously.
Testing protocols have added these scenarios progressively, which is why older ratings do not reflect current capability.
What the rating does not cover
Reliability, which is unrelated.
Long-term structural integrity after repair, which is a genuine concern with modern high-strength steels and adhesive bonding.
Rear seat occupant protection, which has received less attention historically and is now being addressed as testing recognised that rear seat outcomes had not improved at the same rate.
And child restraint installation ease, which is assessed separately in some protocols.
What actually matters practically
Recent test year, since protocols tighten.
Whether active safety systems are standard or optional, since ratings sometimes reflect optional equipment.
And the specific scores rather than the star total, since a vehicle can reach five stars with a weak result in one area offset by others.
The detailed reports are published and are considerably more informative than the headline figure.
Real-world data
The complement to crash testing and arguably more informative.
Insurance and government databases record actual collisions, injuries and vehicle types, allowing comparison of real outcomes.
Which captures effects that testing cannot — how vehicles perform in the collision types that actually occur, at real speeds, with real occupants.
These analyses have shown that some vehicles performing well in tests perform less well in the field, and vice versa.
They are published in several markets and receive far less attention than star ratings.
Repairability
An emerging concern that affects safety indirectly.
Modern structures use high-strength steels and adhesive bonding that require specific repair procedures and equipment.
Repairs not following manufacturer procedures can leave a structure that does not perform as designed in a subsequent collision.
Which is invisible to a subsequent owner and is a genuine reason to obtain repair history when buying used.
Driver assistance limits
Systems are assessed under defined conditions and perform variably outside them.
Testing has found substantial differences in performance at night, in poor weather, and with pedestrians versus vehicles.
Which means the presence of a system says less than its measured performance, and the detailed results are published.
Child occupant protection
Assessed as a separate component in most protocols and frequently overlooked.
It covers performance with child restraints installed and the ease of installing them correctly.
Incorrect installation is extremely common in observational studies, which means ease of correct fitting has real safety consequences beyond the crash performance itself.
Older vehicles
The safety gap between a vehicle from two decades ago and a current one is substantial and is visible in real-world injury data.
Which is worth weighing when buying used, since a newer smaller vehicle frequently offers better protection than an older larger one despite the mass argument.