Although there is increasing evidence supporting the associations between social capital and health, less is known of potential effects in Latin American countries. Our objective was to examine associations of different components of social capital with self-rated health in Colombia. The study had a cross-sectional design, using data of a survey applied to a nationally representative sample of 3025 respondents, conducted in 2004–2005. Stratified random sampling was performed, based on town size, urban/rural origin, age, and sex. Examined indicators of social capital were interpersonal trust, reciprocity, associational membership, non-electoral political participation, civic activities and volunteering. Principal components analysis including different indicators of social capital distinguished three components: structural-formal (associational membership and non-electoral political participation), structural-informal (civic activities and volunteering) and cognitive (interpersonal trust and reciprocity). Multilevel analyses showed no significant variations of self-rated health at the regional level. After adjusting for sociodemographic covariates, interpersonal trust was statistically significantly associated with lower odds of poor/fair health, as well as the cognitive social capital component. Members of farmers/agricultural or gender-related groups had higher odds of poor/fair health, respectively. Excluding these groups, however, associational membership was associated with lower odds of poor/fair health. Likewise, in Colombians with educational attainment higher than high school, reciprocity was associated with lower odds of fair/poor health. Nevertheless, among rural respondents non-electoral political participation was associated with worse health. In conclusion, cognitive social capital and associational membership were related to better health, and could represent important notions for health promotion. Human rights violations related to political violence and gender based discrimination may explain adverse associations with health. ► The good: social capital is relevant given positive effects of trust and political participation on self-rated health. ► The bad: low socioeconomic status people reported worse health. Civic participation was low despite legal mandates. ► The ugly: members of gender and peasant organizations exhibit worse health, potentially due to violations of human rights.