When estimating earnings equations for men in the United States, a dichotomous variable for whether or not the man is currently married is often included as a regressor. The coefficient estimate for this variable is most usually large and significant. However, there is rarely much discussion of the marriage effect. This effect is central to this study, which contributes to the understanding of this statistical association in two ways. First, it shows that the relationship exists in almost all of the fourteen developed countries examined and across several different time periods. Controlling for age, and, when available, education, race/ethnicity, hours worked, and location, marriage differences in annual earnings in favor of currently married males range from 0 percent to 30 percent. Second, it finds that there are important differences between those who are separated, divorced, widowed, and never married.