Few previous studies have analyzed the association between different dimensions of social capital and mental disorders. This study examines whether there is an association between a relatively new theoretical concept describing the amount of trust between individuals and societal institutions, i.e. linking social capital, and hospitalization due to depression or psychosis. The entire Swedish population aged 25-64, a total of 4.5 million men and women, was followed from January 1, 1997, until the first hospital admission due to depression or psychosis during the study period, or the end of the study on December 31, 1999. Small area neighborhood units were used to define neighborhoods. The definition of linking social capital was based on mean voting participation in each neighborhood unit, categorized in tertiles. Multilevel logistic regression was used to estimate odds ratios and neighborhood-level variance in four different models. The results showed strong associations between linking social capital and hospitalization due to depression or psychosis. These associations decreased considerably in both men and women, but still remained significant, after adjustment for age, housing tenure, education, employment status, marital status, and country of birth. However, the results for depression (both men and women) no longer remained significant after adjustment for neighborhood deprivation. In contrast, the results for psychosis decreased considerably but remained significant after adjustment for neighborhood deprivation. The association between low levels of voting participation and hospitalization due to depression or psychosis might reflect neighborhood differences in linking social capital, which could affect vulnerable individuals negatively. Decision-makers should take into account the evidence of a neighborhood effect on mental health in decisions regarding the sites of psychiatric clinics and other kinds of community support for psychiatric patients.