What the Crash Data Actually Says About Old Cars vs. New Cars
Pre-1984 cars have roughly 55% occupant fatality rates versus 36% for 2003-2007 models, and the gap keeps widening. If you're buying old for privacy or repair-cost reasons, here's the safety cost you're trading against.
Mirko4 min read
On this page4 sections
This is part of our car privacy and ownership series. The previous posts covered what connected cars collect, what license plate readers track regardless of your car, and why repairs cost more on newer vehicles. This one covers the side of the ledger that tends to get left out of "old cars are better" arguments entirely: crash survivability.
The numbers
NHTSA's crash-data analysis, using its Fatality Analysis Reporting System (FARS), shows occupant fatality rates around 55% in pre-1984 vehicles, dropping to 36% in model-year 2003-2007 vehicles. That's not a small gap, and it isn't a static one either: it keeps widening as more recent model years add features like standard stability control, curtain airbags, and automatic emergency braking. A separate NHTSA study found the fatality-risk gap between men and women drivers, historically higher for women, shrank from about 18% in older vehicles to 2.9% in model-year 2015-2020 vehicles, largely attributed to safety systems designed with more representative crash-test data.
This is a real, measured outcome, not a model or a projection. If a piece of writing tells you cars used to be built better without mentioning this data, it's leaving out the part of the story that would complicate the argument.
What's driving the newer numbers
A lot of the recent gain traces to advanced driver-assistance systems (ADAS): forward collision warning, automatic emergency braking, and pedestrian detection. Forward collision warning alone reduces rear-end crashes by about 23%, according to IIHS research. Combined with automatic braking, that reduction climbs to 50 to 56%. Separately, an AAA Foundation for Traffic Safety study from August 2023 modeled that wider ADAS adoption could prevent roughly 249,400 deaths and 37 million crashes by 2050, though it's worth being clear that figure is a projection built on assumptions about adoption rates and technology improvement, not an outcome anyone has observed yet. The study itself flags substantial uncertainty in either direction.
The NHTSA fatality-rate data and the IIHS crash-reduction percentages are measured, historical facts. The AAA Foundation's 2050 number is a model. Both are useful, but they're different kinds of evidence, and it matters which one you're citing when the conversation gets specific.
The trade-off, stated plainly
If your reason for buying an older car is behavioral-data privacy (covered in this series' first post) or lower repair costs (covered here), you should know you're also accepting a real, measured increase in crash risk. That's not a reason to avoid an older car. It's a reason to make the trade-off knowingly instead of on the strength of "they built things better back then," which isn't what the crash data shows.
It's also worth separating "old" from "poorly built." The 55%-to-36% gap in NHTSA's data spans roughly two decades and multiple generations of safety-standard changes, not a binary of old versus new. A lot of that gain is concentrated in specific technologies that became standard at specific points, which means the honest question isn't "how old is too old," it's "which safety features does this specific car have."
A more useful way to shop
Rather than picking a decade and working backward, it's worth checking what a specific model year and trim actually included. Electronic stability control became federally mandated on new U.S. vehicles starting with the 2012 model year, though many manufacturers included it earlier voluntarily. Curtain side airbags and backup cameras followed similar phase-in patterns, standard on some trims years before they were required across the board. A 2011 or 2012 model with the safety-package trim can carry most of the protection of a much newer car, while a base trim from the same year might not.
This is the same logic behind the "2011 to 2016 sweet spot" mentioned in the overview post: it's not a magic window, it's the years where a meaningful number of models had most of the modern safety technology standard or widely available as an option, while connectivity and cellular telematics were still inconsistent across trims. Checking the build sheet or window sticker for the specific safety options on the car in front of you will tell you more than the model year alone ever will.
