Health literature shows that unemployment has a gendered effect on health. However, whether men or women are more affected and why remains unclear. We assume that unemployment harms women less than men because of two mechanisms: social roles theories and health selection. First, the availability and centrality in individuals’ lives of roles other than employment may reduce the detrimental effect of unemployment for women. Second, the gendered impact of unemployment on health results from the different ways selection mechanisms operate across genders. Moreover, these two mechanisms may operate differently in different contexts – for example, across different gender regimes. We investigate this by pursuing a three-step comparative approach. The analysis relies on EU-SILC data covering Italy and Sweden for 2004 to 2015 and SOEP data for Germany (1995–2017) and applies correlated dynamic random-effects probit models. While we find weak support for the role of health selection in shaping the relations between unemployment, health, and gender, our empirical results are in line with the hypothesis of a larger gendered effect in older (vs younger) cohorts, western (vs eastern) Germany and Italy (vs. Sweden). Future empirical research needs to directly address the role of gender regimes in moderating such complex relationships.