Every new vehicle sold in America goes through a battery of laboratory crash tests before reaching dealerships. The results get published as star ratings, letter grades, and Top Safety Pick awards. Shoppers read them, dealers display them, and insurers use them. The whole system runs on the assumption that a car that performs well in a lab test will perform well on the road.
The system is right more often than it is wrong. But it is wrong often enough that the gap deserves attention. A 2023 Kia Rio and a 2023 Audi Q7 both meet federal safety standards and both passed crash tests to earn their ratings. On actual American roads, over the 2021-2024 period, drivers of Kia Rios died at a rate of 170 per million registered vehicle years. Drivers of Audi Q7s died at a rate of zero. The gap between the lab assumption and the real outcome is roughly the entire story of this article.
This piece covers three things: where lab crash tests reliably predict real-world safety and where they systematically miss, what the actual real-world fatality data shows (including specific named models), and how IIHS and NHTSA are updating their test protocols to close the gaps. It builds on the How to Interpret IIHS and NHTSA Safety Ratings article, which covers the rating systems themselves; this piece focuses on where those systems don’t match reality.
Why Lab Tests Were Built the Way They Were
The laboratory crash test program in the United States dates to the 1970s, with substantial revisions in the 1990s and 2010s. The core idea has always been to create a repeatable, standardized way to evaluate occupant protection in a defined crash scenario. A vehicle crashes into a fixed barrier at a specific speed, instrumented dummies record forces experienced during impact, and engineers assess how well the vehicle protected the dummies’ heads, necks, chests, and legs.
This approach has specific strengths. It is repeatable — different vehicles can be compared against each other using identical conditions. It is actionable — manufacturers know exactly what the test measures and can engineer vehicles to perform well on it. It is transparent — results get published and consumers can use them to make purchase decisions. The improvements in frontal and side-impact occupant protection since the 1990s are substantially attributable to the pressure that lab testing put on manufacturers.
The approach also has structural limitations. The test measures occupant protection in a specific scenario; it does not measure crash avoidance, driver behavior, roadway interactions, or the safety of people outside the vehicle. A car that performs well in a 40-mph frontal offset test may still kill its driver at high rates if the car attracts aggressive drivers who crash at 80 mph. A pickup truck that protects its occupants beautifully may kill passengers in the smaller cars it hits. A vehicle with good lab scores but poor headlights may rack up fatalities in rural nighttime crashes that lab tests never simulate.
Real-world crash data — primarily from the Fatality Analysis Reporting System (FARS) and the Highway Loss Data Institute (HLDI) — captures what actually happens on American roads. IIHS and NHTSA have both increasingly incorporated real-world findings into their lab test designs, but the lag between a new type of real-world crash pattern emerging and a lab test being updated to address it can be years.
The Real-World Picture: Who Lives and Dies by Vehicle Class
The IIHS Driver Death Rates study, most recently updated in 2026 using 2021-2024 fatality data on 2023 model-year vehicles, provides the clearest picture of how vehicle choice actually affects fatality outcomes. The industry-wide average is 37 driver deaths per million registered vehicle years. The spread across classes is dramatic.

Minicars kill their drivers at 158 per million — roughly four times the industry average. Even though many minicars have Good IIHS crash ratings for their size class, the ratings compare them to other minicars, not to the SUVs and pickups they may actually collide with. A minicar meeting its class’s crashworthiness standards is still a minicar, and physics still applies.
Small cars (79), large cars (104), and large sports cars (75) also run well above the industry average. The large-car category is particularly interesting because it includes the Dodge Charger, Chrysler 300, and Dodge Challenger — vehicles with robust crash ratings whose drivers nonetheless die at high rates. IIHS attributes the gap to driver behavior; the Challenger and Charger attract aggressive drivers who crash at higher speeds than the average driver.
Midsize cars (67) sit above the industry average largely because the class includes several models with below-average performance, including some bestsellers.
SUVs dominate the safest end of the distribution. Midsize SUVs average 24 driver deaths per million, large SUVs 26, very large SUVs 20, midsize luxury SUVs 17, and large luxury SUVs just 10. Even luxury pricing cannot fully explain this; the physics of larger, heavier, higher-riding vehicles meaningfully improves occupant outcomes in multi-vehicle crashes.
Pickups show a split profile. Small pickups average 31 driver deaths per million, large pickups 29, and very large pickups 24 — all below the industry average. But the “other driver” death rate (people killed in vehicles hit BY the pickup) tells a different story. Very large pickups average 110 other-driver deaths per million — the highest of any class — while midsize luxury SUVs average 37 and large luxury SUVs average 26. Pickups protect their own occupants exceptionally well and endanger others at the highest rates in the market.
The split between driver and other-driver death rates reveals something lab tests don’t directly measure: vehicle choice is a decision with externalities. Buying a very large pickup lowers your own crash fatality risk but raises everyone else’s. Buying a midsize luxury SUV achieves both goals simultaneously — among the lowest driver death rates and among the lowest other-driver death rates. The How to Interpret IIHS and NHTSA Safety Ratings article covers the broader rating framework; this real-world data adds the dimension lab ratings mostly don’t capture.
The Models That Buck the Pattern
Zooming in from class averages to specific named models reveals the real spread. Several vehicles with acceptable or better crash-test ratings have driver death rates four to five times the industry average. Several others have recorded zero driver deaths across millions of registered vehicle years.

The Kia Rio leads the industry with 170 driver deaths per million registered vehicle years — more than 4.5 times the industry average. The Nissan Versa follows at 164. Both are subcompact cars that have earned acceptable IIHS crashworthiness ratings and five-star NHTSA ratings. They meet all federal safety standards. On actual roads, their occupants die at rates that would make any lab crash test engineer uncomfortable to defend.
The Chrysler 300 2WD (132), Dodge Charger HEMI 2WD (127), and Dodge Challenger 2WD (125) form a performance-car cluster where the issue is clearly driver behavior as much as vehicle design. These vehicles are purchased disproportionately by drivers seeking performance, driven more aggressively, and crashed at higher speeds than the average car.
The Kia Forte (120), Chevrolet Malibu (116), and Hyundai Venue 2WD (113) round out the top of the list — mostly compact and subcompact cars whose structural limitations show up when physics meets a larger vehicle.
At the other end, six models recorded zero driver deaths across their reporting period: the Audi Q7 4WD, BMW X7 4WD, Chevrolet Tahoe 4WD, Mercedes-Benz GLB 4WD, Mercedes-Benz GLE 2WD, and Toyota Tundra CrewMax 4WD. The next tier of extremely low rates — Porsche Macan 4WD (4), BMW X5 4WD (5), Toyota RAV4 Prime 4WD (5), and Volkswagen Atlas 2WD (5) — is dominated by midsize to large SUVs.
Three patterns emerge from this data that lab tests either can’t fully capture or capture only indirectly. First, size and weight matter enormously for crash outcomes. The physics of a 5,500-pound SUV meeting a 2,500-pound sedan does not favor the sedan’s occupants, regardless of the sedan’s crashworthiness rating. Second, driver self-selection matters. Vehicles that attract aggressive drivers show worse outcomes than their structural ratings predict. Third, luxury models disproportionately populate the safest lists — IIHS’s own analysis suggests driver characteristics (age, income, risk profile) may explain part of the gap, not just vehicle design.
The Externality: Vehicles That Kill Other Drivers
The least-discussed aspect of vehicle safety is what happens to people in the vehicles you hit. Lab crash tests measure occupant protection almost exclusively; they do not measure crash compatibility — whether a specific vehicle is more or less dangerous to the other party in a two-vehicle collision. Real-world data makes this blind spot visible.

The Ram 3500 crew cab long bed 4WD kills other drivers at 204 per million registered vehicle years — nearly four times the industry average of 53. The Dodge Charger HEMI 2WD (161) and Dodge Charger 2WD (136) make the list through sheer mass and speed. The rest of the top 11 reads as a who’s-who of heavy-duty pickups: Chevrolet Silverado 3500 crew cab 4WD (126), Ram 1500 Classic crew cab short bed 4WD (124), Ram 1500 Quad 2WD (113), Ram 2500 crew cab short bed 4WD (113), Chevrolet Silverado 2500 crew cab 4WD (111), Ram 2500 crew cab long bed 4WD (102), Ford F-150 SuperCrew 2WD (99), GMC Sierra 2500 crew cab 4WD (98).
Ten of the top 20 most-dangerous-to-others vehicles are large or very large pickups. All earn acceptable or better crashworthiness ratings. None get penalized by the lab rating systems for the harm they do to other road users.
The lowest other-driver death rates, by contrast, go to midsize to large luxury SUVs with modern collision-mitigation technology: BMW X3 2WD (7), Volvo XC40 4WD (8), Mercedes-Benz GLS 4WD (9), BMW X7 4WD (11), Volkswagen Atlas 4WD (13). These vehicles combine substantial mass (providing some crash compatibility) with aggressive deployment of automatic emergency braking and other ADAS systems that actively avoid collisions.
The pedestrian and cyclist dimension is a related blind spot. US pedestrian deaths rose roughly 60% between 2010 and 2022, a period when crash-test ratings were generally improving. Research has linked much of this increase to the proliferation of large pickups and SUVs with tall, blunt front ends — vehicles that hit pedestrians above the center of mass, pushing them under the vehicle rather than deflecting them onto the hood. IIHS added pedestrian AEB testing in 2019 and tightened it in 2024; NHTSA’s new crashworthiness pedestrian protection program takes effect with the 2026 model year. Both represent lab-test updates trying to catch up with real-world data.
Where Lab Tests Actually Do Predict Reality
The lab-to-road gap is real but not universal. IIHS’s own research has consistently found strong correlation between crashworthiness ratings and real-world fatality outcomes within comparable vehicle classes. Specifically:
Vehicles that earn Good ratings on the IIHS frontal small-overlap test have roughly 50% lower driver death rates in frontal crashes than vehicles rated Poor on the same test, after controlling for vehicle size and driver demographics. The 2012 introduction of the small-overlap test was itself a response to real-world crash data showing that vehicles with Good moderate-overlap ratings were still producing fatal injuries in a specific crash scenario the moderate test didn’t simulate. The small-overlap test addressed that gap, and fatality rates in small-overlap-pattern crashes have declined meaningfully since.
The updated IIHS side impact test (introduced 2021) similarly addressed a real-world pattern: vehicles passing the original side test were still producing fatal injuries when struck by larger, heavier SUVs. The updated test uses a 4,200-pound barrier at 37 mph — closer to the energy of a modern SUV-vs-car collision — and has driven substantial side-impact structural upgrades in vehicles that have since been redesigned.
Pedestrian AEB ratings correlate with real-world reductions in pedestrian injuries, and vehicle-to-vehicle AEB (tested at 31, 37, and 43 mph as of 2024) correlates with substantial reductions in rear-end collision fatalities.
In other words: within comparable vehicle sizes, crash-test ratings do predict real-world outcomes. The breakdown is mostly across vehicle classes, where a Good-rated minicar still can’t match the real-world performance of an Acceptable-rated large SUV, and in behavioral factors that lab tests cannot measure.
How Testing Protocols Are Catching Up
Both IIHS and NHTSA have been updating their testing protocols to address real-world data gaps, though at different paces and with different approaches.
IIHS updates since 2020: – Side impact test (2021): Upgraded barrier weight and speed to simulate modern SUV impacts on passenger cars. – Moderate overlap front test revision (2022): Added rear-seat passenger dummy to evaluate back-seat protection; several vehicles with Good front-seat ratings earned Marginal or Poor rear-seat ratings on first testing. – Pedestrian AEB (tightened 2024): New test for pedestrian detection and braking in multiple scenarios, including darkness. – Vehicle-to-vehicle AEB (2024): Replaced earlier simpler rear-end AEB test with three-speed, three-target protocol testing against passenger cars, motorcycles, and semitrailers. – Headlight evaluation requirements (expanded 2020): Must apply to all trims of a vehicle, not just the top trim. – 2026 Top Safety Pick criteria tightening: Raised the floor on moderate overlap test (Good required for both awards), added V2V AEB requirement for TSP+.
NHTSA NCAP updates (effective with 2026 model year): – Pedestrian automatic emergency braking testing – Lane keeping assist evaluation – Blind spot warning and blind spot intervention evaluation – New crashworthiness pedestrian protection program – Tightened performance criteria for existing AEB testing
The Tesla Model Y was the first vehicle to pass NHTSA’s updated eight-part ADAS safety benchmark under the 2026 protocols. The updated NCAP represents the biggest change to the federal rating program since the 2011 revisions.
Still not directly tested by either program, even in 2026: – Crash compatibility (how dangerous a vehicle is to others it collides with) – Vehicle weight’s effect on pedestrian survival – Long-term reliability of ADAS systems as they age – Real-world nighttime visibility beyond basic headlight evaluation – Driver behavior interactions with specific vehicles
These remaining gaps are where the next decade of safety rating evolution will likely focus. The IIHS has published research on crash compatibility suggesting that very large pickups impose roughly $600 per year in externalized safety costs per vehicle on the rest of the road — a figure that would support regulatory action on crash compatibility if political will emerged. It has not yet.
What This Means for Shoppers
The practical implications for a shopper deciding between vehicles are specific and sometimes counterintuitive:
A five-star NHTSA rating or IIHS Top Safety Pick+ means the vehicle is well-designed for its class. It does not mean the vehicle is safe in absolute terms if the class itself has high real-world fatality rates. A five-star minicar is still a minicar.
For occupant safety specifically, size matters as much as rating. A midsize or large SUV with acceptable ratings will typically produce better real-world fatality outcomes than a minicar or small car with good ratings, because the physics of larger vehicle mass and higher ride height work in the occupants’ favor in multi-vehicle crashes.
For the lowest driver fatality risk, midsize and larger luxury SUVs dominate the data. Audi Q7, BMW X5 and X7, Mercedes-Benz GLE and GLS, Volvo XC90, Toyota Highlander, and similar vehicles consistently show low to zero driver deaths across millions of vehicle years. The reasons are a combination of vehicle design, advanced safety technology deployed earlier at luxury price points, and driver self-selection.
For road-user overall safety (your own plus others), midsize SUVs and smaller luxury SUVs are the sweet spot. Large and very large pickups protect their drivers but impose significant fatality costs on others. Buyers who value overall road safety rather than just their own occupant protection should weigh this.
Avoid specific models with documented high driver death rates. The Kia Rio, Nissan Versa, Chrysler 300, Dodge Charger HEMI, Dodge Challenger, Kia Forte, Chevrolet Malibu, and Hyundai Venue appear repeatedly in the worst-performing lists despite acceptable or better crash ratings. Some of this is driver self-selection, but some is structural.
Modern ADAS features matter disproportionately. Automatic emergency braking (both pedestrian and vehicle-to-vehicle), lane-keeping assist, and blind-spot monitoring have produced measurable real-world fatality reductions — effects that don’t appear in traditional crashworthiness ratings. A vehicle with excellent ADAS and acceptable crashworthiness often outperforms a vehicle with great crashworthiness and poor ADAS in real-world outcomes.
Headlight quality matters more than most shoppers realize. IIHS added headlight ratings in 2016 after finding that some luxury vehicles’ standard halogen headlights performed worse than much cheaper vehicles’ LED systems in real-world visibility. Rural nighttime crashes are a significant source of fatalities, and headlight quality disproportionately affects this outcome.
The Should You Trust Tech in First-Year Car Models article covers the question of how safety tech performs in its first production year, and the Which Factory-Order Options Are Worth It piece discusses which safety packages reliably deliver value in the purchase process.
The Limits of All This Data
The real-world data has its own limitations that are worth acknowledging. Driver death rates are influenced by who buys and drives each vehicle, not just the vehicle itself. The Dodge Challenger’s high driver death rate reflects both the vehicle’s handling characteristics and the fact that aggressive drivers disproportionately buy Challengers. If a Challenger were somehow driven only by cautious retirees, its fatality rate would be much lower.
Fatality data also reflects miles driven, annual mileage, type of driving, and geographic distribution — none of which the IIHS rate calculations fully control for. A vehicle driven primarily in urban stop-and-go traffic produces different fatality statistics than the same vehicle driven primarily on rural interstates.
The industry average of 37 driver deaths per million registered vehicle years is substantially better than it was 20 years ago. Overall US traffic fatality rates per mile have declined significantly over the long term, though they’ve plateaued or risen slightly in recent years. The improvements are partly attributable to better vehicle design, partly to better infrastructure, and partly to better ADAS technology — all driven in meaningful part by the pressure of both lab testing and real-world data analysis.
Crash avoidance technology has introduced a new wrinkle: a crash that doesn’t happen doesn’t show up in any fatality statistic. The best measure of a modern vehicle’s safety may not be how it performs when crashed but how often it avoids crashing in the first place. Lab tests have begun to measure this through AEB protocols, but the real-world benefit of advanced driver-assistance systems is harder to isolate from other variables.
The Takeaway
Lab crash tests remain a valuable tool, but their relationship to real-world fatality outcomes is more complicated than star ratings suggest. The ratings work well for comparing vehicles within the same size class and identifying structural strengths and weaknesses. They work less well for predicting absolute fatality outcomes across classes, where vehicle size, driver behavior, and crash compatibility often dominate.
The IIHS real-world fatality data for 2023 model year vehicles shows a 17x spread between the safest and most dangerous vehicles in driver death rate — from zero deaths per million registered vehicle years in several luxury SUVs to 170 per million in the Kia Rio. The spread in other-driver death rates is even wider, with very large pickups killing other drivers at 4x the industry average. Neither of these dimensions is captured in a star rating or a Top Safety Pick badge.
The practical approach for a safety-conscious shopper combines both data sources: use lab ratings to identify well-designed vehicles within whatever size class fits your needs, and use real-world fatality data to inform the class choice itself. A midsize or larger SUV with modern ADAS and good lab ratings is the clearest path to both low driver death rate and low other-driver death rate. For buyers whose budget or lifestyle requires a smaller vehicle, the lab ratings are the best available guide, with the understanding that physics still applies to a small car in a world full of heavy pickups.
Both IIHS and NHTSA are actively updating their testing to close the gaps the real-world data has exposed — pedestrian AEB, improved side impact, V2V AEB, rear-passenger protection, and new ADAS evaluation protocols are all responses to what fatality data was showing that lab tests were missing. The protocols will continue evolving, but will always lag the real-world patterns somewhat. Treating both the lab and the road data as complementary, rather than either alone as the complete story, is the right approach.
External Sources
- IIHS — Driver Death Rates Remain High for Small Cars
- Insurify — Supersize Pickups, Muscle Cars, and Small Cars Are Deadliest Vehicles
- IIHS — As Heavy EVs Proliferate, Their Weight May Be a Drag on Safety
- IIHS — Supersizing Vehicles Offers Minimal Safety Benefits but Substantial Dangers
- Repairer Driven News — New IIHS Report on Smaller, More Powerful Vehicles
- Cars.com — Death Rates Highest for Small Cars, Models With Powerful Engines
- CarPro — IIHS Driver Death Rates Report
- KBB — Study: Small Cars Top Deadliest List, SUVs Least Deadly
- Stacker — 30 SUVs With Highest Death Rates
- NHTSA — Fatality Analysis Reporting System (FARS)
- IIHS — Relationships of Frontal Offset Crash Test Results to Real-World Driver Fatality Rates
- Informed for Life — 2026 Car Safety Report
- Carscoops — IIHS Death Rate Study 2026































































Leave a Reply
You must be logged in to post a comment.