Baseline data on the perceived health status of participants (N = 5,201) in the Cardiovascular Health Study of the Elderly (CHS) are reported. The authors examined the predictive utility of health-related factors representing eight different domains, assessed gender differences in the prediction of perceived health, and tested a hypothesis regarding the role of known clinical conditions versus subclinical disease in predicting perceived health. Multivariate analyses showed that the majority of the explained variance in self-assessed health is accounted for by variables that fall into four general categories. Although gender differences were small, the analysis showed that the relative importance of several predictor variables did vary by gender.