98% of Recalled Cars Are Fine. That's Why Nobody Fixes Them.

Dodge recalled 48,777 Hornets and Alfa Romeo Tonales this month for rear seat belts that could twist and fail to retract. But Stellantis disclosed something in the NHTSA filing: an estimated 1.6% of those vehicles actually have the defect. Just 780 cars, while the other 47,997 are fine.

Dodge is not alone here: in June, Hyundai recalled 96,310 Tucson variants for instrument panels that could blank out, and NHTSA estimated about 1% were affected.

NHTSA's 2025 Annual Recalls Report logged 670 campaigns covering 28 million vehicles with a 50.3% completion rate.^[1] Everyone blames lazy owners, backordered parts, lost mail. But when the average manufacturing recall has a 1-to-2% defect rate, the system is a screening test with a catastrophic false-positive rate, training its own users to stop complying.

Consider how this plays out: you get a recall letter, schedule a dealer visit, surrender your Tuesday morning. All clear, the call says, your vehicle did not have the defect. This happens once and you shrug it off; twice and you resent it; by the third letter, you toss it in the recycling. In diagnostic medicine, a screening test with a 98% false-positive rate would be pulled from clinical use. Mammography accumulates roughly 50% false positives over ten annual rounds, and even that measurably reduces future compliance.^[2]

Completion data tracks this with uncomfortable precision: 83% for vehicles under three years old, 60% by years five through ten, 52% by eleven through fifteen, 38% past sixteen.^[3] An owner of a ten-year-old car has received half a dozen recall notices, and if most led to wasted trips, the rational response is to stop going.

Manufacturers cannot pre-screen which VINs carry a manufacturing defect. When a bolt was not torqued correctly on a random Tuesday and fifty thousand vehicles came off the line that quarter, the only defensible strategy is recalling them all, even though the system then does the safe thing by recalling everyone, which makes the safe thing look like a false alarm, which makes fewer people comply next time, which leaves the 780 genuinely defective vehicles buried in a population that has learned to throw recall letters away.

Two caveats. NHTSA does not require manufacturers to disclose defect rates for all campaigns, so the 1-to-2% figure draws on a small sample. And software, fire-risk, and structural recalls typically affect far higher percentages of the population; the false-positive dynamic applies to manufacturing-line defects, which generate the majority of campaigns by volume.

VIN-level component traceability exists in aerospace and is technically feasible for automotive; some OEMs already track engine blocks and transmissions at the serial-number level. Extending it to seat belt retractors and inverter bolts would cost money, but the next recall letter you receive would be accurate, and you might actually listen to it. In the meantime, check your VIN at nhtsa.gov/recalls. If you have an open recall, take it in; the system only works if people show up, even when the odds say your car is fine.