Every Car Under $20,000 Is Dead. The Replacements Are Deadlier.

Before you start browsing used car lots because the sticker on every new car reads like a mortgage payment, consider this: zero new cars are available under $20,000 in 2026, down from 17 options just six years ago. The Nissan Versa, the last holdout of America's cheapest-car club, was discontinued in December 2025,^[1] and the cheapest new car you can buy right now is a Kia K4 at $22,290.

Good riddance? Those cheap cars killed plenty of people, sure, but what's replacing them in budget buyers' driveways is worse.

Nine Headstones

We tallied the FARS body count across every sub-$20K car that's been axed since 2020: Nissan Versa (722 deaths, 0.90 per 100M VMT), Chevy Spark (517 deaths, 1.28 rate), Ford Fiesta (513, 1.02), Chevy Sonic (494, 1.40), Honda Fit (290, 0.72), Kia Rio (268, 1.07), Toyota Yaris (190, 0.76), Chevy Aveo (138, 0.69), and the Mitsubishi Mirage (81, 0.40).^[2] That's 3,213 deaths spread across a decade of FARS data, with an average rate of about 0.91 per 100 million vehicle miles traveled and an average lethality hovering around 70% when a crash turned fatal and involved one of these cars.

Most of these were objectively dangerous cars, and the Chevy Sonic's 1.40 rate paired with a 75.4% lethality should embarrass GM's engineering heritage. But the Mirage, with a 0.40 rate and 65.3% lethality, sits on par with the Kia Forte, which serves as the basis for the K4, the very car that now occupies the cheapest-new-car throne. Mitsubishi killed a car that was genuinely doing its job for people who needed a $16,000 vehicle.

Where Budget Buyers Land Now

Average new car transaction price in January 2026: $49,191, per Cox Automotive,^[3] which means a buyer who could afford $16,000 on a new Mirage isn't stretching to $49K and is instead buying used. The share of used vehicles available under $20,000 has collapsed from 53% in 2019 to 30% in 2025, according to J.D. Power,^[4] and those remaining sub-$20K used vehicles are older, with median ages pushing 8-12 years.

FARS model-year data quantifies how much that matters: vehicles from model year 2018 accumulated 4,435 deaths during the 2014-2023 data window, model year 2013 vehicles racked up 6,903, and model year 2005 hit 11,363.^[2] Some of that gap reflects longer exposure time for older vehicles, but it also reflects the absence of electronic stability control (mandated for all vehicles only after 2012), automatic emergency braking, and the structural improvements that have dropped per-mile fatality rates for newer model years steadily over two decades.^[5]

Run the math: the average dead-cheap-car rate was 0.91 per 100M VMT, while the Kia Forte, the cheapest viable new alternative, manages 0.40, a 2.3x gap that flatters the dead cars because the replacement scenario is even worse. Budget buyers who are now forced into a 2014-2016 used vehicle are likely facing fatality rates somewhere north of 1.0, based on the model-year death curve, meaning they traded a 0.91 risk floor for something closer to the 1.2-1.4 range that characterized the worst cars on this list.

What You Should Do

If $22,290 for a new K4 is within reach, stretch. Its fatality rate is half the average of the cars it replaced, and IIHS gave it a 2026 TOP SAFETY PICK+.^[6] If you can't afford new, prioritize used vehicles from model year 2018 or later, when AEB availability became widespread and post-ESC-mandate designs had matured through at least one major refresh cycle. Before signing anything, run your VIN at nhtsa.gov/recalls for open recalls, because the cheap-used segment has the lowest recall completion rates in the industry.

Strongest Counterargument

Cheap new cars were killing people, and a Chevy Sonic with a 75.4% lethality rate and a 1.40 fatality rate per 100M VMT deserved to be pulled from production. A five-year-old used Honda Civic, even with higher exposure-adjusted risk, has modern airbags, ESC, and structural rigidity that a new Spark never matched, so for specific used-car-to-dead-cheap-car comparisons the safety argument can swing either direction. The problem is that the median buyer priced out of $22K isn't getting a five-year-old Civic; they're getting whatever's cheapest on the lot, which increasingly means a 10-year-old vehicle without the safety features that separate survivable crashes from fatal ones.

Limitations

FARS only captures fatal crashes, which are a fraction of the roughly 6.7 million total annual crashes. A vehicle with a low fatality rate might still injure occupants at high rates. Our estimated_rate figures use VMT estimates rather than actual odometer readings, introducing roughly ±15% uncertainty for low-volume models like the Mirage (81 deaths over a decade is a thin dataset). The model-year death curve reflects cumulative exposure over the 2014-2023 window, so older model years had more time on the road; direct rate-per-mile comparisons across model years require fleet-size normalization we haven't performed here. Finally, this analysis is prospective: we're predicting that budget buyers displaced from new sub-$20K cars will land in older, more dangerous used vehicles, but we can't yet measure that displacement quantitatively.

Data sourced from NHTSA FARS 2014–2023, IIHS, Cox Automotive, J.D. Power, and manufacturer pricing data; estimated fatality rates use VMT approximations with ±15% uncertainty for low-volume models, and readers should consult the methodology section for additional caveats on rate calculations and fleet size estimates.