The present study was designed to provide information on the relationship between self-reports of health and physicians' ratings in an aged sample, and to determine how both of these measures of health relate to longevity. Subjects were 69 survivors (median age = 84.25 years) of a sample of aged twins who had been followed longitudinally since 1947 to 1949. Self-reports of health were found to be significantly correlated with ratings assigned by a physician on the basis of medical records. Both types of measures were predictive of differences in survival time among the younger subjects in the sample, but neither was significantly related to longevity for older subjects. The results suggest that self-reports could provide a valid, cost-effective means of health assessment in studies in which other forms of health information are lacking.