What the evidence says
Patrols with automated license plate readers (ALPRs) produce more detection of stolen vehicles and auto thieves than patrols without them. Lum (2025) summarizes the Mesa, Arizona randomized experiment this way: patrols with and without ALPRs both produced short-term reductions in crime at the patrolled locations, and detection of stolen vehicles and auto thieves was higher when ALPRs were used. The mechanism is detection: the reader flags plates of stolen vehicles that officers would not otherwise check. Lum notes that deterrence from an ALPR-equipped car may be present even when the reader is off, so the experiment cannot cleanly separate deterrence from ALPR-specific detection.
For stolen vehicle recovery, the direction is consistent across sources, but the evidence is mostly associational. Kelsay (2024) reports that prior studies associate ALPR technology with increased vehicle recoveries. He cites those studies without giving their designs, sample sizes or effect sizes, so no magnitude comes from that source. Lum (2025) reports that Koper et al. (2022) found ALPRs may increase stolen vehicle recoveries.
Arrests are less clear. Koper et al. (2022), as summarized by Lum, found that ALPRs may increase recoveries but not necessarily arrests of people stealing cars. Kelsay (2024) and Lum (2025) also describe studies that associate ALPRs with more arrests. Ozer (2016) found more follow-up arrests per month for violence and property crimes after ALPR implementation in Cincinnati, with unclear effects on deterrence. These arrest findings are not specific to auto theft.
Investigative use is a smaller part of the picture. When ALPRs were used in investigations, about 46% of documented uses were to locate stolen tags or vehicles. Case clearance occurred in about 19% of cases in which ALPRs were used, and those cases were about 1% of the cases examined. For almost three-quarters of the sample where ALPR use was noted, ALPRs did not contribute in any meaningful way to the resolution of the investigation.
On auto crime itself, the evidence differs by design. Lum's summary of randomized experiments found that thirty-minute ALPR patrols, conducted once every few days in randomly selected high-crime places for two to three months, did not reduce auto-related or other crime compared with control locations. A 2026 event study of fixed ALPR deployment (Mourtgos) found that motor vehicle theft fell 11.0 percent after deployment (95 percent CI [-17.3, -4.2]). Adoption in that study is not random.
Where the evidence splits
Short patrol experiments and fixed deployment disagree on auto crime. The randomized patrol experiments found no reduction in auto-related crime. The event study of fixed deployment found an 11.0 percent decline in motor vehicle theft. The sources differ in design: the experiments used thirty-minute patrols every few days for two to three months, while the event study used fixed deployment with staggered adoption. The sources on this page do not say whether dosage, deployment type or identification strategy explains the split. The claims note that the experimental dose may have been too low to detect an effect, and that adoption in the event study is non-random. Neither point is tested here.
Arrest findings also differ. Koper et al. (2022) found no clear increase in auto theft arrests, while Kelsay (2024) cites several studies associating ALPRs with increased arrests, and Ozer (2016) found more follow-up arrests in Cincinnati. The sources do not say whether the difference comes from the crime type (auto theft versus violence and property crime generally) or from study design. They do not reconcile it.