The civilian labor market performance of women veterans is examined, using both traditional data sets and a special survey of reservists that facilitates control for selection by the military and enlistees. Evidence from all data sets indicates that female veterans possess a higher level of measured earnings endowments than do nonveterans. Moreover, evidence from the reservist survey supports the hypothesis that unobserved ability and preferences toward the military are positively correlated with measured endowments. Absent control for unmeasured quality and preferences, the observed wage advantage of female veterans is accounted for entirely by differences in measured characteristics. Following their control, a wage disadvantage is found for female veterans, relative to their nonveteran counterparts. An exception is among nonwhite female veterans, who realize wage rates similar to nonveterans. The low returns to military service for women may result from the narrow opportunities for skill enhancement given women within the military, and an inability to transfer these skills to the civilian sector owing to the specificity of the training or occupational barriers facing civilian women.