The driving performance of 20 subjects with central vision impairment due to either Stargardt disease or cone-rod dystrophy (visual acuity, 20/40 to 20/70) was compared with that of 29 control subjects with normal vision who had similar driving histories. Driving performance was defined by accident involvement based on self-report and state records and by an evaluation of performance on an interactive driving simulator. The proportion of individuals involved in accidents in the central vision loss group was comparable to that of the control group. For 13 of the 20 subjects with central vision loss who did not restrict their driving to daylight hours, there was a greater likelihood of involvement in nighttime accidents than in the control group. Visual function measures and simulator indexes did not predict accident involvement for the central visual loss group, although these subjects showed longer braking response times and a greater number of lane boundary crossings than the control group. These findings are in contrast to our previously published report of subjects with retinitis pigmentosa, who were more likely to have been involved in both daytime and nighttime accidents than a control group and for whom visual field extent was significantly related to accident involvement.