Recent studies have investigated the relationship between social capital and health, although relatively few studies have evaluated the role of gender in this relationship. This study's aim was to investigate whether individual-level social capital in communities is associated with the self-reported health (self-rated health and depression) of urban elderly women and men living at home, even after adjusting for some confounding factors, and whether there is a gender difference in the association between social capital and health. A self-administered questionnaire was mailed to 2,400 people aged between 65 and 79 who were not receiving long-term care insurance services in City A (Tokyo). Indicators of social capital, as represented by cognitive social capital (trust in neighbors and reciprocity) and structural social capital (group membership in local associations), were tested with two self-reported health indicators: perceived health and depression. The associations between social capital and perceptions of health were analyzed for men and women respectively using multivariate logistic regression analysis with adjustments made for age, economic status, educational status, presence of chronic disorders, functional capability levels, marital status, and duration of residence. The number of valid responses was 1,776 (men: n=887, 71.2 ± 4.0 years; women: n=889, 70.9 ± 3.9 years; mean ± SD) with a response rate of 74.5%. Lower levels of cognitive social capital (civic mistrust) were associated with an odds ratio for poorer self-rated health of 1.58 (95% CI [1.07-2.34], P=.022) at the individual level in men. In women, lack of reciprocity (OR=1.63, 95% CI [1.10-2.41], P=.014) was associated with poorer self-rated health. Civic mistrust and lack of reciprocity were associated with depression in men and women. Lack of group membership in local associations was associated with self-rated health (OR=1.68, 95% CI [1.16-2.44], P=.007) and depression (OR=2.24, 95% CI [1.49-3.38], P<.001) in women. Civic mistrust was associated with poorer self-rated health and a lack of reciprocity was associated with depression in men. In women, lack of reciprocity was associated with poorer self-rated health and depression, and lack of group membership in local associations was associated with self-rated health and depression. These results lead to the conclusion that there were noticeable gender differences in the relationship between the social capital and self-rated health of the elderly. A longitudinal study should be conducted to clarify the causal relationship between social capital and perceived health.