Background: More knowledge is needed about how various dimensions of social capital associate with mortality either independently or mediated by other factors. In this cohort study, individual-level social capital was assessed as a possible predictor of all-cause mortality. Methods: A 9-year mortality follow-up was conducted for participants of a nationally representative Health 2000 Survey carried out in 2000−01. The survey included a personal interview, self-administered questionnaires and a health examination. Response rates to different items varied between 80 and 88%. Due to the item non-response, missing values were replaced by using multiple imputation. Participants (N = 6377) were 30–79 years old at the baseline. During the follow-up period, 245 women and 340 men died. Cox regression models were used to assess the association between social capital and mortality. We analysed three dimensions of social capital: social support, social participation and trust. Their association with the risk of death was adjusted for several socio-demographic factors, health behaviours, health and biological risk factors. Results: Unactive social participation was associated with higher mortality risk even when adjusting for the other factors. The association was stronger among men [hazard ratio (HR) = 1.56] than women (HR = 1.42). The other two dimensions of social capital were associated with mortality when only age was adjusted for, but this association disappeared when social participation was included in the explanatory model. Conclusions: Active social participation protects against early death. This association is partly mediated through health behaviour and to a minor extent also through health.