In later life, social ties become increasingly important for one’s mental health. In this study, we test whether exogamous unions, in which one of the partner is migrant and one native, affect older individual’s social networks, and whether the association between exogamy and mental health operates through social networks. We hypothesize that immigrants gain social capital through their native spouse, and thus have better mental health than endogamous immigrants. In contrast, the kin network of exogamous natives is not expanded through their spouse, and we therefore expect their mental health to be lower than that of their endogamous counterparts. We investigate these hypotheses with data from the German Socio-Economic Panel Study. In line with our hypotheses, we observe that exogamous immigrants have larger social networks due to the expansion of their kin network through their native spouse, while exogamous natives have smaller social networks than endogamous natives. We find that natives who marry immigrants exhibit worse mental health than endogamous natives, while exogamous immigrants’ mental health is better than that of endogamous migrants. Yet, most of these differences can be attributed to selection effects. Once we account for unobserved heterogeneity using correlated random effects models, we find that exogamous native women exhibit significantly worse mental health than any other group. Social networks influence mental health, but contribute little to these observed differences, suggesting that unobservable characteristics attached to those who form exogamous unions are at play. This work enriches the discussion on the effects of social networks and the diversification of union types on mental health in older ages.