GM Built the Deadliest Sedan in America. Then the Safest. Then They Killed It.

General Motors spent three decades building compact sedans in America, and the trajectory reads like a confession written in crash data. The Cavalier killed at 2.43 per 100 million vehicle miles traveled. The Cobalt, its replacement, managed 5.10 — more than double, one of the highest death rates of any sedan ever mass-produced in this country.^[1] Then came the Cruze, and something genuinely strange happened.

It worked. The Cruze posts a rate of 0.63, which makes it the fourth-safest sedan in the entire FARS database among models with 400 or more recorded deaths, safer than the Camry, safer than the Civic, safer than the Accord and the Corolla and the Malibu and every other sedan that outsold it by a comfortable margin. The only cars with lower rates are the Kia Forte, the Prius, and the Kia Optima, which is the kind of company that would have given a GM marketing executive a stroke in 2011.

And it is not because the Cruze attracted more cautious drivers who stayed sober behind the wheel and coasted through school zones at 22 miles per hour. The impairment numbers are nearly identical across all three generations: 22.4% for the Cavalier, 22.4% for the Cobalt, 20.2% for the Cruze.^[1] Same demographic buying the cheapest sedan on the Chevy lot, same proportion getting behind the wheel impaired, radically different outcomes in the morgue.

What changed was the car itself, not the person driving it. The Cavalier and the Cobalt shared the GM Delta platform, a design philosophy that treated occupant protection as an afterthought stapled to a cost target. The Cobalt’s ignition switch, the one that would shut off the engine, disable the airbags, and kill at least 124 people before GM admitted it knew,^[2] was the most famous defect, but not the only structural problem. The 2006 model year alone accounts for 328 deaths in the FARS data. That is one model year of one sedan from one manufacturer producing a body count that exceeds some entire vehicle nameplates across their full production run.

The Cruze arrived in 2011 on GM’s global Delta II platform, co-developed with Opel, engineered to meet European crash standards that were, at the time, meaningfully stricter than American ones. It had standard electronic stability control from day one — a technology that the Cobalt never received because GM stopped updating that platform the moment the ignition switch lawsuits hit.^[3] Structurally, the Cruze is a different animal: modern crumple zones, a stiffer passenger cell, better side-impact geometry, the full catalog of improvements that separate a car designed in 2009 from one designed in 1999.

The result showed up immediately in the data. The Cruze’s first full model year in FARS logged 49 deaths, while the Cobalt’s first full model year logged 169. By 2019, the Cruze’s final vintage, annual deaths had dropped to 34, less than one-tenth of the Cobalt’s worst year.

GM discontinued the Cruze in March 2019 and closed the Lordstown, Ohio plant that built it. The stated reason was the same reason every automaker gives: Americans buy trucks and SUVs now, not sedans. But what GM actually did was take the first vehicle in its compact sedan history that wasn’t actively contributing to a public health crisis and scrapped it, because the profit margin on a Silverado is five times higher.

They have since killed the Sonic, the Volt, the Impala, and the Malibu. GM currently sells zero sedans in the United States. The company that spent twenty years learning how not to kill people in a small car finally learned, and then decided the lesson was no longer commercially interesting.

The strongest case against this reading

Newer cars are safer than older cars, full stop, and the Cruze is newer than the Cobalt by a decade. Some portion of that 8.1x improvement is just calendar years doing their work: better materials science, better crash testing, better regulatory standards that every manufacturer met. The Cobalt’s rate is inflated by the ignition switch defect, which was a singular manufacturing failure rather than a systemic engineering posture. Controlling for that defect alone might halve the gap, though it would not close it.

What to do with this

If you are shopping for a used sedan under $15,000: the 2014–2019 Cruze is statistically one of the safest cars you can buy at any price point, and it is priced like what it is, a discontinued Chevy that nobody wants. The 2017 model year is the sweet spot: the second-generation refresh with better infotainment, the same safety bones, and a used-market price that suggests the buying public has no idea what the FARS data says. Check your VIN at nhtsa.gov/recalls regardless; the Cruze had its share of recalls, including one for brake booster failure.

Limitations

FARS captures only fatal crashes; VMT estimates from the NHTS carry ±15% uncertainty for lower-volume models. The Cavalier data includes pre-2000 model years when crash survival standards were broadly worse, inflating its rate relative to what a Cavalier-era equivalent would look like on modern roads. Impairment percentages reflect only fatally crashed drivers, not overall driving populations. The Cobalt’s 5.10 rate includes ignition-switch deaths, which were a defect, not normal crash performance. Removing those deaths would lower its rate but not bring it close to the Cruze.