The hypothesis that periodic motor vehicle inspection (PMVI) has no safety effect was tested using accident involvement rates analysed by "vehicle age" and "time since the most recent inspection." The alternative of interest was that the probability of accident is lowest (ceteris paribus) immediately after an inspection, and subsequently increases over time. Two types of adjustment for exposure variations by time since last inspection were made, yielding two kinds of accident involvement rate. The first accident rate was the proportion of accident-involved vehicles having "preventable" defects which could possibly have helped to cause the accident. The second accident rate was the number of accident-involved vehicles divided by the number of inspected vehicles, and amounted to an adjustment for premature re-inspection. The observed probability of accident involvement (as measured by either rate) was found to increase with time since last inspection. This result supports the alternative hypothesis that a mandatory safety inspection has an immediate safety benefit which decreases over time. In neither analysis was there an interaction between vehicle age group and "week since inspection."