Empirical evidence showed that married men generally earn more and married women earn less. However, the control group of “not married” differs between studies, over time and between countries, so that the message remains somewhat fuzzy. It is not clear whether the type of union or being in union is responsible for these wage penalties and premium. This article aims to analyse whether the marriage pay more than cohabitation in a country where both unmarried and married partnerships coexist for years, such as France. Thanks to a rich dataset with information on both marital and work history of both partners, we are able to estimate the effect on hourly wage of being married relatively to being in a consensual union. Taking into account selection into marriage and on the labour market and possible differential in specialization level, our results show that the men’s marriage premium is entirely due to positive selection into marriage. While marital specialization process within couple strongly decrease women’s hourly wage, there is no evidence of an additional marriage penalty for women. The gender wage gap between married partners or unmarried partners is similar, once controlled for selection into marriage.