One Investigation Per 2,008 Complaints: Inside NHTSA's Defect Pipeline

NHTSA's Office of Defects Investigation received 76,304 consumer complaints over the last four fiscal quarters and opened exactly 38 investigations.^[1] That ratio works out to one investigation for every 2,008 Vehicle Owner Questionnaires. Put another way, 99.5% of consumer safety complaints go into the filing cabinet and stay there.

The quarterly metrics tell a story about scale that gets worse the longer you stare at it. In the most recent quarter alone, FY26 Q2, NHTSA took in 20,262 complaints and opened nine investigations. Those nine investigations triggered recalls covering 12.6 million vehicles, meaning each investigation averaged 1.4 million affected units. A decade ago, the DOT Inspector General found the same office running with a defect investigations budget that had been frozen at $10 million for nearly ten years, staffed with investigators the OIG described as "insufficient or ill-trained" for the complexity of modern vehicles.^[2] The complaints have only gotten more numerous since then, while the vehicles have gotten exponentially more complex.

Consider what's sitting in that pipeline right now. On August 3, NHTSA upgraded its investigation into Ford's 1.0-liter EcoBoost engine to an engineering analysis covering 135,551 Fiesta, Focus, and EcoSport vehicles from 2014 through 2021. The agency logged 355 complaints alleging sudden power loss from timing belt degradation. Ninety-eight percent of failures happened before the 150,000-mile maintenance interval Ford prescribed. Average failure mileage: 70,000 miles.^[3] Ford discontinued all three models. Convenient timing, given that FARS data shows the Focus accumulated 3,046 deaths during the period the agency spent in preliminary evaluation, running at a fatality rate of 2.52 per 100 million vehicle miles traveled.^[4]

The Ford probe is not an anomaly. Last October, NHTSA expanded its investigation into 286,000 GM vehicles with the L87 6.2-liter V8 after 1,157 engine bearing failure reports. The Chevrolet Silverado 1500, one of the affected models, leads the entire FARS database with 9,591 deaths across the dataset period.^[5] In April, the Range Rover steering knuckle investigation widened to 331,559 vehicles after Jaguar Land Rover's own recall covered only 2015 through 2017 models while fractures kept appearing on 2020 and 2021 units.^[6] And just last week, FCA recalled 1.27 million Ram 1500 trucks because the second-row seat buckle anchors were never properly attached to the body structure.^[7]

Each of these cases followed the same trajectory: preliminary evaluation, years of additional complaints, then an upgrade to engineering analysis or a recall. The GAO flagged this pattern in 2016, noting that NHTSA "faces several challenges in its oversight of vehicle safety defects" including inadequate data analysis, insufficient staff guidance, and an inability to keep pace with new technology.^[8] The agency's response was to build a new IT system called the Corporate Information Factory. The complaints kept arriving, but the investigations stayed flat.

The uncomfortable math runs like this: 39.6 million vehicles were recalled in the last four quarters. About 280 million registered vehicles sit on American roads. Roughly one in seven was recalled in a single year. The system catches the biggest defects eventually, and the recall populations prove it. But "eventually" is doing a lot of heavy lifting when the pipeline processes fewer than one-half of one percent of the signals consumers send it, and the vehicles involved in active investigations have been killing people in FARS data for years before the investigation even opens.

What you should actually do: File a VOQ at nhtsa.gov/report-a-problem every time something safety-relevant happens with your vehicle, even if it feels minor. Volume is how complaints cross the screening threshold. Check your VIN at nhtsa.gov/recalls quarterly, not annually. And when NHTSA issues a "do not drive" order for your vehicle, take them literally. That language is reserved for cases where they've calculated the failure probability is high enough to kill you before the dealership schedules you in.