Married, cohabiting, and divorced men in Sweden earn more than single men. The wage premium earned by married men has declined since 1968, mainly due to decreasing productivity differences between married and single men. During this period, reforms have been undertaken to induce spouses to share labor market and housework more equally. If this wage differential reflects specialization within households, we would expect it to decline. Using longitudinal data, the results indicate that the wage premiums mainly reflect gains from partnership. Selection based on unobserved productivity into partnership can only partly explain the wage differentials by marital status. However, I do not find that the marriage premium increases with time married as also implied by the specialization hypothesis.