What percentage of car seats are installed incorrectly?

[NHTSA](https://www.nhtsa.gov/car-seats-and-booster-seats/keep-kids-safe-road) puts it at almost half. You will see much more precise figures quoted elsewhere, often broken down by seat type to the percentage point, and those breakdowns are not published by NHTSA or the CDC. Treat "roughly half" as the honest number. The errors technicians see most often are a loose belt path, the wrong recline angle, the wrong harness slot, and a missing top tether on forward-facing seats. A free check with a certified technician is the reliable way to find out which, if any, apply to your install.

How much do car seats actually reduce injury risk?

In passenger cars, [NHTSA](https://www.nhtsa.gov/car-seats-and-booster-seats/keep-kids-safe-road) reports that car seats reduce the risk of fatal injury by 71% for infants and by 54% for toddlers aged 1 to 4, compared with being unrestrained. Belt-positioning boosters reduce the risk of serious injury by 45% for children aged 4 to 8 compared with a seat belt alone. Those three figures are the best-established numbers in child passenger safety, and they assume the seat is installed and used correctly.

How many children are killed or injured in crashes each year?

In 2023, 1,019 child passengers aged 14 and younger were killed in motor vehicle crashes in the United States and more than 161,000 were injured, according to [NHTSA](https://www.nhtsa.gov/car-seats-and-booster-seats/keep-kids-safe-road). Of the children who died, 43% were not buckled up. That last figure is the one worth sitting with: it is not a statement about which seat you buy, it is a statement about whether a child was restrained at all.

Car Seat Statistics 2026: Usage & Safety Data

Q: How many lives do car seats save?

Among children under 5, [NHTSA](https://www.nhtsa.gov/sites/nhtsa.gov/files/protectingchildren.pdf) estimated 325 lives saved in 2017 by correct restraint use, and calculated that a further 46 children would have survived had every child been properly buckled. Since 1975, child restraints have saved more than 4,000 children from fatal injuries. Figures circulating for lives saved across all children under 13, or for injuries prevented annually, are not published by NHTSA and should be treated with suspicion.

Q: Are all children equally likely to be buckled up?

No, and the gap is one of the clearest findings in the data. A 2021 observational study cited by the [CDC](https://www.cdc.gov/child-passenger-safety/risk-factors/index.html) found 21% of Black children and 15% of Hispanic children aged 4 to 7 were not buckled up, against 7% of White children. Restraint use also falls as children get older: among child passengers aged 8 to 12 in 2021 fatal crashes where restraint status was known, 43% were unrestrained, against 27% of those aged 0 to 3.

Most car seat statistics you will read online are invented. Not exaggerated or out of date, but fabricated outright and then attributed to NHTSA, because a made-up number with a federal agency’s name attached reads as authoritative and almost nobody checks.

So this page is deliberately short. Every figure below links to the agency that published it. Where a number is commonly quoted but has no published source, that’s said plainly rather than filled in.

How much difference a car seat makes

These three figures are the best-established in child passenger safety, and they come from NHTSA. All compare correct restraint use in passenger cars against the alternative.

Restraint	Reduction in risk	Compared with
Rear-facing car seat, infants	71% risk of fatal injury	Unrestrained
Forward-facing car seat, ages 1-4	54% risk of fatal injury	Unrestrained
Belt-positioning booster, ages 4-8	45% risk of serious injury	Seat belt alone

Two things about that table matter more than the numbers. The first is that every figure assumes the seat is installed and used correctly, which is a much larger assumption than it sounds. The second is that the booster row is measured against a seat belt rather than against nothing, which is why moving a child out of a booster too early is a real decision with a measurable cost, not a milestone.

Lives saved, and the ones that weren’t

Among children under 5, NHTSA estimated 325 lives saved in 2017 through correct restraint use, and calculated that a further 46 children would have survived had every child been properly buckled. Since 1975, child restraints have saved more than 4,000 children from fatal injuries.

That second number is the useful one. It is the cost of the gap between what we know and what we do.

Who is dying

In 2023, 1,019 child passengers aged 14 and younger were killed in motor vehicle crashes in the United States, and more than 161,000 were injured, per NHTSA.

Of the children who were killed, 43% were not buckled up.

That figure is not about choosing between a good seat and a better one. It is about whether a child was restrained at all, and it accounts for a larger share of child deaths than any question of seat selection or installation quality.

How many children are actually buckled

A nationally representative observational survey cited by the CDC found 89% restraint use for ages 4 to 7 and 87% for ages 8 to 12.

Restraint use falls as children get older, and the fatal crash data shows the same pattern more starkly. Among child passengers where restraint status was known in 2021 fatal crashes, 43% of those aged 8 to 12 were unrestrained, against 27% of those aged 0 to 3.

Across all child passengers aged 12 and younger in 2022 fatal crashes where status was known, 35% were unrestrained.

Parents buckle babies. The habit erodes somewhere around the age a child starts objecting, which is also the age at which a booster still does real work.

The gap nobody advertises

Restraint use is not evenly distributed. A 2021 observational study cited by the CDC found 21% of Black children and 15% of Hispanic children aged 4 to 7 were not buckled up, against 7% of White children.

Those differences track access to free inspection services, work schedules that make a weekday fitting appointment impossible, and the cost of replacing a seat a child has outgrown. They are not differences in how much parents care. If you are looking for where child passenger safety effort produces the most benefit per hour, it is here, and a certified technician check is free.

Misuse: the number everyone quotes

NHTSA says almost half of car seats are installed incorrectly.

That is the whole published figure. You will find far more precise versions in circulation, broken down by seat type, with error categories carrying their own percentages. Those breakdowns are not published by NHTSA or the CDC, and this page used to repeat several of them before they were checked.

What technicians consistently report seeing, without attaching invented percentages to it:

A belt path with more than an inch of movement, the single most common installation error
The wrong recline angle, which matters most for newborns
Harness straps in the wrong slot for the direction the seat is facing
A missing top tether on a forward-facing seat, which is also the most commonly skipped step in the LATCH system
A chest clip sitting on the belly rather than at armpit level

Where the child sits

The CDC recommends age- and size-appropriate restraints in the back seat until adult seat belts fit properly, which it puts at age 8 or at least 57 inches (145 cm) tall, whichever comes later in practice.

Note that this is a restraint-fit threshold rather than a front-seat threshold. The American Academy of Pediatrics separately recommends that children stay in the back seat until 13, and those two recommendations answer different questions. A child can pass the belt-fit test years before the back seat stops being the safer place to sit. Our guide on when a child can sit in the front seat works through the difference.

What these numbers don’t tell you

Three limitations are worth carrying, because they are rarely stated alongside the figures.

The effectiveness percentages are for passenger cars. NHTSA reports separate, generally lower figures for light trucks, and the 71% and 54% quoted everywhere are the passenger car values. If you drive an SUV or pickup, the direction of the benefit is the same but the magnitude isn’t the one on the poster.

Observational surveys count restraints, not correct use. When a survey reports 89% restraint use among 4 to 7 year olds, it is recording that a restraint was present and in use. It is not recording whether the harness was at the right height, whether the belt path was tight, or whether the child had outgrown the seat. Set that against NHTSA’s “almost half are installed incorrectly” and the two figures describe different things. Both can be true at once, and together they are more sobering than either alone.

Restraint status is frequently unknown. The fatal crash percentages above are calculated only from cases where investigators recorded whether the child was restrained. Cases with unknown status are excluded rather than assumed, which is the right methodology, but it means the real unrestrained share could sit either side of the published figure.

None of this makes the numbers unusable. It makes them numbers rather than verdicts, which is how any safety statistic deserves to be read.

What the numbers actually ask of you

The data points at three things, in order of how much they change the odds.

Buckle the child, every trip. 43% of the children who died in 2023 were unrestrained. Nothing else on this page moves the number that much.

Keep them in the right stage longer. Rear-facing outperforms forward-facing, and a booster outperforms a seat belt by 45% for a 4 to 8 year old. The pressure is always to move up early.

Get the install checked once. If almost half of seats are wrong and you have never had yours inspected, the base rate says it’s worth an hour. Check your state’s requirements too, since legal minimums generally sit below what the evidence supports.

Reading car seat statistics elsewhere

Since this page exists to be a reference, one practical warning. Fabricated child safety statistics are common, and they share recognisable tells:

Suspicious precision. “59% of seats are installed incorrectly” reads as more authoritative than “almost half”, which is exactly why it gets invented.
Breakdowns that no agency collects. Error categories with percentages attached, or usage rates split by income bracket, are almost never real.
A named agency with no link. “According to NHTSA” followed by no citation is the most common form. NHTSA publishes its data; if a figure is real, it can be linked.
A study that cannot be found. A named journal and year with no DOI or title is worth five minutes of searching before you repeat it.