I Ranked Every Brand by Generational Safety Gains. Kia Finished Last.

The methodology is straightforward enough that I am slightly embarrassed nobody published it before. Take FARS fatal crash data from 2014 through 2023. Group vehicles into two generation windows by model year: old (2005–2012) and new (2013–2022). Calculate annual deaths per model-year cohort for each manufacturer. Compare.

Lincoln topped the chart at negative 75.4%, dropping from 54 deaths per model year to 13. Acura, Chrysler, GMC, Chevrolet, and BMW all cleared 55% improvement, Ford and Honda landed around 50%, Toyota managed 43.9%, and even Nissan, which I expected to stall, pulled off 25.1%. Hyundai and Jeep barely moved, scraping by with single-digit improvements. VW and Subaru went slightly backwards, at plus 1.0% and plus 16.4%, respectively.^[1]

Kia: 103 deaths per model year in the old generation, 204 in the new, plus 98.8%, the sole brand out of sixteen to go backwards.

Now for the honest part

Kia's U.S. sales grew from 356,268 units in 2010 to 782,451 in 2023, a 2.2-times increase that roughly mirrors the 2.0-times death increase.^[2] The per-vehicle fatality rate is approximately flat. The individual model data confirms this: a Kia Sportage posts 0.28 deaths per 100 million VMT, a Sorento 0.29, and a Telluride 0.04, all competitive with or better than their class averages.^[1]

So Kia's cars did not get more dangerous; they got more numerous, and that distinction matters enormously if you are buying a Kia, because your individual risk is fine, and it matters not at all if you are measuring whether America's fleet got safer, because the aggregate body count still doubled while everyone else was cutting theirs in half.

Why stagnation is worse than it sounds

Toyota's U.S. sales also grew during this period, as did Hyundai's, Honda's, and Ford's. Toyota grew by roughly 20% and still cut its generational death toll by 43.9%, while Honda grew and cut 49.1%. The industry invested in electronic stability control (mandated for all new vehicles in 2012), better side-impact structures, advanced airbag geometries, and automatic emergency braking that actually works in real-world collisions rather than just on test tracks, and these investments yielded 40% to 60% fatality reductions even as fleets expanded.^[3]

Kia adopted the same technologies: ESC became standard on its vehicles by the 2012 mandate date and IIHS ratings improved across the lineup. But Kia grew so fast that the safety dividend was entirely consumed by volume. The Optima is the clearest example: 17.6 deaths per model year in the old generation, 57.6 in the new, a 227% increase driven almost entirely by fleet penetration rather than any engineering regression.^[1]

The strongest case against this finding

Deaths per model year is confounded by fleet size, and FARS does not publish per-model registration data that would allow normalization. Kia happened to be the fastest-growing major brand in America during the study window. A brand that triples its fleet would triple its deaths at a constant rate. Comparing raw death counts across brands with wildly different growth trajectories conflates safety engineering with market success. A fair test would divide deaths by registered vehicle-years per model-year cohort. That data exists at state DMV level but not in FARS, and I do not have it.

What you should actually do

If you are shopping for a Kia, the brand-level trend should not scare you away. The Telluride's 0.04 per-100M-VMT death rate is among the lowest in any class. The Sportage and Sorento compete with the RAV4 and CR-V on per-mile safety. Buy on model, not on brand.

If you already own a Kia Optima (now K5), Soul, or Forte from model years 2013 through 2018, check your VIN at nhtsa.gov/recalls. Kia recalled 3.37 million vehicles between 2018 and 2023 for engine fire risks across multiple model lines, and recall completion rates for these campaigns have been lower than the industry average.^[4]

Limitations

FARS captures fatal crashes only. The roughly 36,000 annual deaths represent a sliver of the 6.7 million total crashes per year. A brand could have improving injury rates and worsening fatality rates, or vice versa, and this analysis would not detect it. Model-year cohort death counts do not control for vehicle age, mileage, driver demographics, geography, or road type. Newer model years have more accumulated exposure in the 2014–2023 observation window, which biases newer-generation counts upward for all brands, though this confound applies uniformly and does not explain why Kia is the sole outlier. Per-VMT estimates use nationally averaged mileage assumptions from the NHTS rather than actual odometer readings, introducing approximately plus or minus 15% uncertainty for low-volume models.^[5]