Independently done surveys of a target population can make an important contribution to knowledge about the determinants of personal health behavior by highlighting variables that consistently emerge as significant predictors. This investigation examined the correlates of four health practice and knowledge indices related to cardiovascular disease (CVD) in two baseline community surveys of the Pawtucket Heart Health Program (N = 2,413; N = 2,808). An additional dimension was the use of three adult age groups (18-29, 30-49, 50-64) in conducting the analyses. Results of both surveys showed that sex was the strongest correlate of the four indices--knowledge of CVD, encouraging health practice changes in others, dietary intake, and exercise. The four indices related to CVD were also associated with years of education, primary language, and whether or not a recent cholesterol measurement had been obtained, although these relationships were not as consistent as the results for sex. Overall, about half of each survey's significant associations were also found in the other survey (survey 1, 30 of 62; survey 2, 30 of 56). Consistency of significant results between surveys was best for the group ages 30-49. In either survey, it was rare for an association between a predictor and behavioral index to appear in each of the three age groups. This study supports the importance of the subjects' sex in research on personal health practices, suggests the potential for independence even among health-related indices pertinent to a single type of illness, and emphasizes the usefulness of utilizing independent samples to identify important correlates of health behavior.