Articles / From Complaint to Recall: How Long the G...

From Complaint to Recall: How Long the Government Actually Takes

September 4, 2026 · DefectFile Data Desk

Somewhere between an angry owner filing a complaint and a manufacturer mailing a recall notice sits the federal defect investigation: the formal process where the safety agency decides whether a pattern in the data is actually a defect. The register DefectFile indexes holds 3,522 of them, dating back to the 1980s, each with the date it opened, the date it closed, and the recall campaign it produced, if any. Which means the machinery can be timed with its own paperwork. So: how long does it take?

The median case: about five months

Across 3,421 investigations with usable opening and closing dates, the median ran 143 days, roughly five months, from opening to closure. The average is 246 days, dragged upward by a long tail of multi-year cases, and one investigation in ten ran past 522 days. Closure, importantly, is not the same as a recall: only about a third of all investigations on file, 1,191 of 3,522, are linked in the register to at least one recall campaign. The other two thirds closed without one, which can mean the agency found no defect trend, or that the manufacturer fixed the problem through a service campaign or warranty extension instead. An investigation is a question, not a verdict.

When the answer is a recall, it usually comes within a year

For the 995 investigations where both the opening date and the resulting recall's filing date are on record, the median gap is 120 days. Most of the pipeline moves briskly: 365 of those recalls were filed within 90 days of the investigation opening, and 86 percent arrived within a year. The lag has barely moved in forty years, holding between 105 and 144 days in every decade since the 1980s. Whatever else has changed about cars, the distance from formal question to formal answer has stayed remarkably constant.

The extremes are where the stories live. The fastest conversion in the file took 2 days: an investigation into cracked wheels opened in May 2005 and the recall was filed before the week was out. The slowest took 4.7 years, an anti-lock brake investigation from 1994 that did not produce its recall until 1999. More recently, a 2016 investigation into brake master cylinder failures ran 3.7 years before the recall of 292,311 vehicles was filed in 2020, and a 2011 power steering investigation took 4.4 years to become a recall of 86,018.

Investigations are getting slower, and more decisive

Time the investigations by the decade they opened and a clear shift appears. Cases opened in the 1980s, 1990s, and 2000s all closed in a median of 132 to 136 days. Cases opened in the 2010s took a median of 214 days. Cases opened in the 2020s that have closed so far took 390, and that figure is a floor, not an estimate, because the slowest cases from this decade are still running and have not yet been counted.

At the same time, investigations became far more likely to end in a recall. In the 1980s, 17 percent of investigations on file link to a recall campaign. In the 1990s, 29 percent. In the 2000s, 43 percent, and in the 2010s, 47 percent. Read together, the two trends describe an agency that opens fewer speculative cases and works the ones it opens harder and longer, on problems that have also gotten harder: a sticking throttle cable is one kind of investigation, and driver-assistance software that misreads a parked fire truck is another.

The marathons

The longest completed investigations in the file each ran about seven years. An investigation into intermediate steering shaft separation opened in February 2013 and closed in April 2020. One into vehicles rolling away after being parked opened in December 2016 and closed in January 2024. One into panoramic sunroofs shattering opened in 2014 and ran six and two thirds years. These are the contested cases, where the defect is intermittent, the population is huge, or the manufacturer disputes the pattern, and they are the reason the average investigation lasts months longer than the median one.

The investigations that never close

As of the August 2026 snapshot, the register lists 99 investigations as open. They are two very different groups wearing one label. Sixty-two opened in the 2020s: that is the live docket, the cases actually being worked. But 28 date to the 1990s, including an inquiry into rollover risk opened in August 1996 and a rear compartment lid investigation from earlier that same year, both listed as open for three decades. Nobody is plausibly still working a 1996 trunk lid case; these are almost certainly records that were never administratively closed. The register remembers everything, including the paperwork nobody finished.

How to read an open investigation on your own vehicle

Every open investigation appears on the recall boards and on the affected vehicles' pages, above the recalls, because it is the earlier signal. The base rates in this article are the honest guide to reading one: most investigations close without a recall, roughly a third convert, and when one does convert it typically happens within a year of opening. An open investigation into a vehicle you own or are about to buy is not a verdict. It is the government formally asking the same question the complaint file has been whispering, and the answer, either way, will land on the same page.

Methodology. All figures are from an August 2026 snapshot of the federal defect investigation file maintained by the National Highway Traffic Safety Administration, covering 3,522 investigations. Durations use each investigation's earliest opening date and latest closing date as recorded; 3,421 investigations had usable dates. An investigation is counted as producing a recall when the register itself records a recall campaign number against it, and the investigation-to-recall lag is measured to the filing date of the earliest linked campaign with one on record (995 investigations); the absence of a linked campaign does not prove no related recall ever occurred. Decade medians for the 2020s cover only cases already closed, so they understate that decade's final figure. Investigations listed as open are those with no closing date in the file, which for the oldest entries likely reflects records never administratively closed rather than active work. Counts are of investigations, not vehicles.