The offenders were compared with control groups of drivers matched for age and sex. The study groups included 1,350 Oregon offenders and 742 California offenders who attended the panels and the same numbers of controls. In addition, conviction rates before and after the victim impact panels were studied in 683 drivers who were ordered to attend the panels but failed to do so, as well as the same number of controls. Results revealed that although in Oregon the panel attendees had a lower rate of recidivism than their matched control group, the recidivism rate of the attendees was not different than that of either those ordered but not attending the panels or the control group for the no-shows. In California, no differences were observed in recidivism rates between the panel group and either the no-show group or the two control groups. More focused analyses of specific age and gender groups indicated that the victim impact panel may be an effective tool for offenders over age 35. Findings indicated that in general the victim impact panel is not a very effective method of modifying the behavior of all drunk driving offenders. Tables and 17 references (Author abstract modified)