The Toyota Solara Has the Soberest Drivers in FARS — and a Death Rate 4x the Fleet Average

The Toyota Solara has a 4.1% impairment rate among its fatal-crash drivers — the lowest of any vehicle in the FARS database. Nearly 96% of the people dying in Solaras are stone sober. The car’s death rate? 4.25 per 100 million VMT. Almost four times the fleet average.^[1]

I wanted to know whether this pattern was unique or widespread. So I merged FARS_BY_MODEL with FARS_TOXICOLOGY across 200 vehicles with 50 or more drivers and 100 or more fatal crashes in the 2014–2023 dataset. Normalized each vehicle’s death rate and impairment rate to z-scores against the fleet mean. Subtracted one from the other.^[1]

What falls out is a single number per vehicle that answers a question NHTSA has never explicitly asked: is this car killing people because of how it was built, or because of who drives it? Positive gap = the vehicle is deadlier than its drivers explain. Negative gap = the vehicle’s engineering absorbs its drivers’ worst decisions.

Design Kills

Five vehicles emerged with z-score gaps above +5.0. Each one has a death rate wildly disproportionate to its impairment profile.

The Solara is the case study that should keep Toyota’s engineers up at night. A 4.1% impairment rate across 195 fatal-crash drivers is the lowest in the entire dataset. These were the most responsible drivers in the FARS database. They still died at 4x the fleet average.^[1]

Why? The Solara was a Camry coupe produced from 1998 to 2008. Same platform, less structure. The coupe body eliminated the B-pillar and rear door openings that the sedan used as side-impact load paths. Toyota marketed it as a sportier Camry; structurally, it was a Camry with fewer ways to dissipate energy in a T-bone. Most Solara model years shipped without side curtain airbags — a feature the Camry sedan began offering in 2007 and that IIHS research has shown reduces driver death risk in near-side crashes by approximately 37%.^[3]

The Tracker is a different failure mode. GM’s rebadged Suzuki Sidekick weighed 2,600 pounds and sat on a narrow track with a high center of gravity — a combination that earned it a two-star NHTSA rollover resistance rating in early assessments.^[2] Impairment rate of 12.7%, well below the fleet average of 20.2%. The drivers weren’t the variable. The geometry was.

Driver Kills

The flip side: vehicles where the engineering absorbed the human variable.

The Buick Park Avenue is remarkable. Nearly a third of its fatal-crash drivers were impaired — the highest rate of any vehicle with 100+ drivers in the dataset. Its death rate? 0.48 per 100M VMT. Less than half the fleet average. The car absorbed enough abuse from its own drivers to suppress the fatality rate below the mean.^[1]

At nearly 4,000 pounds on a full-size unibody platform, the Park Avenue carried the structural mass that IIHS research consistently associates with reduced occupant fatality risk.^[4] GM’s late-90s luxury engineering — triple door seals, reinforced rockers, full-length side rails — gave occupants a buffer that cheaper platforms didn’t. The people inside made terrible decisions. The car compensated.

The Policy Fork

Two different failure modes demand two different interventions.

For “design kills” vehicles — the Solara, Tracker, Veloster — ignition interlocks and sobriety checkpoints are treating a symptom that barely exists. The drivers are already sober. What would actually move the needle: accelerated scrappage programs targeting vehicles with high design-kill gaps, inspection regimes that penalize the absence of basic safety tech, and NCAP ratings that stop giving four stars to cars lacking side curtain airbags.^[5]

For “driver kills” vehicles — the Park Avenue, C/K, Astro — the car is fine. The intervention is behavioral: ignition interlocks, ride-share subsidies in DUI-heavy demographics, alcohol treatment programs. Redesigning the Park Avenue wouldn’t have saved anyone. It was already doing its job.

Current policy treats all vehicle fatalities as the same problem. This data says they’re two different problems with two different solutions. Blending them into a single “road safety” budget is like treating a broken bone and the flu with the same prescription.

Methodology

Z-scores computed as (vehicle value − fleet mean) / fleet standard deviation for both death rates per 100M VMT (mean: 1.10, SD: 1.39) and impairment rates (mean: 20.2%, SD: 5.8%). Gap = rate_z − impairment_z. Z-scores make both distributions dimensionless and comparable, but the comparison assumes that one standard deviation of death rate variation and one standard deviation of impairment variation are meaningfully equivalent — an assumption rather than a derivation. Fleet limited to vehicles with ≥50 drivers in FARS toxicology data and ≥100 total fatal crashes across 2014–2023. VMT estimates derived from NHTS fleet size × average annual mileage. Both datasets sourced from the same FARS bulk extraction underlying this site’s other 97 articles.

Limitations

FARS captures fatal crashes only — approximately 38,000 per year out of 6.7 million total. The z-score gap identifies anomalies in fatality patterns but cannot prove causation. A vehicle appearing in the “design kills” column might have a confounding variable — rural usage patterns, highway-speed exposure, or demographic age distribution — that the model doesn’t isolate. VMT denominators use fleet-level estimates with ±15% uncertainty for low-volume models. The impairment variable captures BAC > 0 or drug-positive toxicology among drivers in fatal crashes, not the general driving population, which inflates impairment rates relative to on-road reality. The Solara’s 195 fatal-crash driver sample, while above our threshold, remains small enough that a handful of misclassified toxicology reports could shift its impairment rate meaningfully.

The Counterargument, at Full Strength

Vehicle age is the elephant this analysis doesn’t fully address. Every “design kills” vehicle on the list is old — the Tracker (1999–2004), Solara (1998–2008), Seville (1998–2004). Old vehicles accumulate more miles of exposure, attract lower-income buyers with fewer repair resources, and predate safety mandates. The z-score gap might be measuring poverty and vehicle age rather than “design failure” per se. The Veloster (2012–2019) partially rebuts this — it’s recent enough to have ESC, front airbags, and ABS, yet still carries a gap of +7.09 — but the broader pattern is real. The distinction between “bad design” and “old design that was acceptable at the time” matters for policy, and this analysis doesn’t cleanly separate them.