Helmet Safety Standards Explained

Heads up: we earn a commission if you buy through the links on this page, at no extra cost to you. It never decides what makes the list — here’s how that works.

This page is about head injury, so before anything else, here is how it was built. Every number below was read off a document we fetched on 17 July 2026 and linked at the bottom. Nothing is quoted from memory. Where we tried to fetch a source and failed — and we failed three times, including twice on the CPSC’s own website — we say so, and we leave the gap open rather than fill it with something plausible.

We are not engineers and we do not test helmets. What follows is the regulation, in the regulation’s own numbers, with the marketing subtracted.

CPSC 16 CFR 1203: the American floor

In the United States, bicycle helmets are not certified by a voluntary badge scheme. They are governed by federal law. 16 CFR Part 1203 — the Safety Standard for Bicycle Helmets — is mandatory, and it applies to every bicycle helmet manufactured after 10 March 1999. A helmet that does not comply cannot legally be sold. There is no premium tier of compliance and no budget tier: there is the standard, and there is illegal.

Here is what a helmet has to survive.

Impact attenuation — 300 g.A headform wearing the helmet is dropped onto an anvil, and an accelerometer at the headform’s centre of gravity records the peak. The criterion, at §1203.12(d)(2): “The peak acceleration of any impact shall not exceed 300 g.” Any single impact over 300 g fails the helmet. The test procedure lives at §1203.17.

Drop velocities.Flat anvil: 6.2 m/s ±3%, which needs roughly a two-metre drop. Hemispherical and curbstone anvils: 4.8 m/s ±3%, roughly 1.2 m. The anvils themselves are specified — flat (125 mm minimum diameter), hemispherical (48 ±1 mm radius), curbstone (105° angle, 15 ±0.5 mm edge radius).

Four conditioning environments, because a helmet has to work on the worst day as well as a nice one. Ambient (17–27 °C); low temperature (−17 to −13 °C); high temperature (47–53 °C); and water immersion — each conditioned for 4 to 24 hours before impact.

Positional stability and retention. The helmet must not come off the headform in the roll-off test (§1203.12(b)), and the retention system must hold without elongating more than 30 mm (§1203.12(c)). The Bicycle Helmet Safety Institute describes the yank test as a 4 kg weight dropped 0.6 m onto the strap.

Peripheral vision — 105°.Per §1203.14, vision must not be obstructed within 105 degrees of each side of the midsagittal plane. A helmet you can’t see out of causes the crash it was meant to survive.

Labelling — §1203.34.The compliance label must be durable and must carry the statement “Complies with U.S. CPSC Safety Standard for Bicycle Helmets”, plus the manufacturer or importer, a production lot identification, and an uncoded month and year of manufacture. That last detail is quietly useful to you: it means you can read the actual age of a helmet off the label rather than decode it.

EN 1078: the European standard, and the 250 vs 300 trap

Europe’s standard is EN 1078. Per SATRA, a notified body that tests to it, the shock absorption limit is 250 g (2453 m/s²)— measurably lower than the CPSC’s 300 g. This is where nearly every article about helmet standards stops, concludes “EN 1078 is stricter”, and moves on.

Look at the drop heights before you accept that.EN 1078’s flat-anvil drop is from approximately 1.5 m (up to 89 J). The CPSC’s flat-anvil drop is 6.2 m/s, needing roughly 2.0 m. So the European standard demands a lower peak acceleration from a less severe impact, and the American standard permits a higher peak acceleration from a more severe one. Those are two different trade-offs, not a ranking. Anyone telling you 250 beats 300 is comparing one column of a two-column problem.

BHSI’s side-by-side comparison gives EN 1078 the edge on the g threshold and on the retention system, and notes the CPSC specifies higher flat-anvil velocities. Where EN 1078 is unambiguously more demanding is retention: it allows no more than 35 mm dynamic and 25 mm residual strap extension, versus the CPSC’s 30 mm, and it adds a requirement the CPSC has no equivalent for — the buckle opening force must not exceed 30 N under a 50 kg load. It also conditions helmets with UV ageing (xenon lamp plus water spray cycles), which the CPSC doesn’t; the CPSC uses water immersion, which EN 1078 doesn’t.

The honest summary: they test different things, in different ways, and neither dominates. If you are buying in the US, this is largely academic — a helmet sold to you legally meets the CPSC standard, and many meet both.

MIPS: a real gap, an unpublished magnitude

Now the important structural point, and it is the one that reframes everything above. Look again at the CPSC test: a uniaxial accelerometerat the headform’s centre of gravity. One axis. Peak linear acceleration. Drop a headform straight down, measure how hard it stops.

Real crashes are not that. Heads hit tarmac at an angle, and the head rotates. Rotational motion is not measured by the CPSC criterion at all — and it is not measured by EN 1078’s 250 g criterion either. Both mandatory standards are, at their core, linear standards. That is not a scandal; it is a floor written in the 1990s that has saved an enormous number of lives. But it is a gap, and it is the gap the entire rotational-system industry exists to fill.

MIPS describes its product as a “low-friction layer that is mounted inside the helmet”, “designed to move slightly in the event of an impact” in order to “help redirect rotational motion away from the head”. It identifies rotational motion as “a common cause of concussions and more severe brain injury in oblique hits to the head”. The mechanism is coherent and it targets exactly the thing the law doesn’t test.

Here is what MIPS does notpublish on the page we retrieved: any reduction percentage, any figure for how many millimetres the layer actually slides, and any test protocol. It does publish this, in capitals: “NO HELMET OR IMPACT PROTECTION SYSTEM CAN PROTECT A USER FROM ALL INJURIES.” It acknowledges “there are limits to the protective capabilities of all helmets”.

We have seen specific slide distances and reduction percentages quoted confidently on other cycling sites. We could not source them, so we will not repeat them. If you see a number attached to MIPS, ask where it came from — the company itself isn’t putting one on its front door, which is interesting in both directions.

Virginia Tech: the only public measurement of the gap

Which leaves one genuinely useful tool. The Virginia Tech Helmet Lab rates helmets one to five stars, with four and five recommended, on a score where lower is betterbecause the underlying figure “estimates the number of concussions the average person would see if they experienced identical impacts”.

The protocol: 24 impact tests per helmet, at six dispersed locations, at both medium and high energies, with impact velocities “selected to reflect common cyclist head impact velocities based on helmet damage replication studies”. Crucially it measures both linear acceleration and rotational velocity— the second of which no mandatory standard covers — and weights each result by how often cyclists actually take that kind of hit. The lab describes its work as “an independent and objective assessment of helmet performance for consumers, free from manufacturer influence”.

We could not retrieve the per-model ratings table. The rating rows load dynamically and we did not get them, so this site does not publish a star rating for any individual helmet — not in our helmet roundup and not here. We could have guessed. Guessing a safety score is precisely the thing that would make this page worse than useless. Go and look your own model up.

What we tried to fetch and couldn’t

In the interest of you being able to audit this page: the CPSC’s own regulations page and its business guidance page both returned 403 Forbidden to us, and the eCFR’s Part 1203 page redirected to a bot wall. We therefore took the regulatory text from govinfo.gov, which is the US Government Publishing Office’s official repository of the Code of Federal Regulations — a primary source of equal standing — and cross-checked the key figures against BHSI independently. Both agree on 300 g, on 6.2 and 4.8 m/s, on the anvil geometry, on the conditioning ranges and on the 30 mm retention limit. Two independent retrievals agreeing is the reason we are comfortable printing those numbers.