Stimulated by the finding (Kawachi et al., 1997) that social capital in communities may mediate the relationship between income inequality and health status, this article describes relationships between individual-level elements of social capital--trust, commitment and identity in the social-psychological dimension; participation in clubs and associations and civic participation in the action dimension--and self-rated health status, before and after controlling for human capital (socioeconomic status measured by income and education), using survey data collected in Saskatchewan, Canada (n = 534, 40% response rate). Income (P = 0.001) and education (P < 0.001) were related to health in the expected directions. Both income (P = 0.002) and education (P = 0.004) were related to health among the elderly; education (P = 0.035) to health among the middle-aged; and neither among the youthful respondents. Frequency of socialization with work-mates (P = 0.019) and attendance at religious services (P = 0.034) had the strongest (and positive) relationships with health of the social engagement questions, even after controlling for human capital, and participation in clubs and associations was positively related to health among the elderly (P = 0.009). But for commitment to one's own personal happiness (P = 0.039), trust, commitment and identification of various kinds were not significantly related to health. Civic participation was also unrelated to health. The main conclusion is that little evidence was found for compositional effects of social capital on health. Secondary findings are that the relationship between SES and health was the same for men and women and strongest among the elderly; that socialization with colleagues from work is relevant and that attendance at religious services and participation in clubs are related to health for the elderly.