A body of empirical research has arisen in labor economics concerning the apparent earnings advantage of married men relative to divorced or never married men. These differences remain even after controls are introduced for education, work experience and other determinants of earnings. Some studies suggest that marriage is a social arrangement conducive to acquisition and specialization of human capital. A related explanation is that men who marry assume significant responsibilities, the result of which is increased work commitment and higher earnings. Alternatively, the selection hypothesis suggests that men who are strong earners are more attractive in the pool of marriageable men, and they are more likely to be selected as marriage partners. One explanation that has received less attention is the potential for employer discrimination, which is present if employers predicate wage offers in part on the basis of marital status. This paper’s principal focus is the discrimination hypothesis. Defining discrimination in the statistical sense of Dennis J. Aigner and Glen G. Cain (1977), it finds evidence for U.S. data suggesting employer discrimination in a sample of young men, particularly among those with less than secondary education. In order to ascertain whether the other causes play contributing roles, the paper presents indirect tests of the productivity and selection hypotheses. It finds that the family responsibility and selection hypotheses contribute to the marriage premium.