We evaluate the effects of the transition from cohabitation to marriage on household domestic and market work hours using a sample of working couples. For this purpose we use the 21 first waves of the German Socio-Economic Panel (GSEOP). We adapt the estimator introduced by Semykina and Wooldridge (2005) to system GMM estimation to account for selection bias in the presence of endogenous regressors. Our results indicate that marriage increases women's specialization in home-based activities and that marriage decreases women's leisure. These effects are robust across specifications.