A computerized test of 22 visual, attentional, perceptual, cognitive and psychomotor abilities was administered to 261 elderly drivers age 62 and above. Approximately half of the sample had been reported to licensing agencies for incidents of unsafe performance and half were free of reported incidents. Significant relationships were found between the presence or absence of incidents and all measured abilities. Cognitive deficits evidenced the strongest association with unsafe performance, visual deficits the least, with the remaining inbetween. The magnitude of relationship is presumed to be a joint function of the prevalence of a deficit and its effect upon driving. Intercorrelations among abilities generally exceeded correlations with performance, indicating the need for caution in interpreting reported relationships between individual abilities and any measure of performance, such as crashes. The authors also caution against interpreting the degree of relationship found in highly selected samples as representing the magnitude of relationship in the population at large.