Large-scale prospective studies of disease development often rely on self-reported data. To assess the accuracy of self-reports of cataract, we compared the self-reports with medical record data obtained from diagnosing ophthalmologists and optometrists for participants in the Physicians' Health Study, a randomized trial of aspirin and beta-carotene among 22,071 male U.S. physicians aged 40-84 years. A report of cataract, defined as a positive response to a question about whether cataract had ever been diagnosed in either eye and the date of diagnosis, was found to be a very good indicator of lens opacification but was not a good indicator of an incident, age-related opacity that reduced visual acuity. These results indicate that in large prospective studies of clinically significant cataract, where examination of all study participants is not feasible and self-reported data are used, additional documentation to supplement the self-reports should be obtained and strict diagnostic criteria applied to minimize the likely effects of misclassification.