The majority of armed forces veterans make use of the subsidized training and educational benefits provided by the Department of Veterans Affairs. The effect of veterans benefits on educational attainment am civilian earnings is estimated here using the Census Bureau's 1987 Survey of Veterans. Two identification strategies are employed to control for unobserved characteristics that are correlated with educational attainment and benefit usage. First, a fixed effects strategy is implemented by exploiting information on educational attainment at the time of entry to service. Second, instrumental variables estimates are computed, where the excluded instruments are interactions between period of service am educational attainment at entry to service. The effect of veterans benefits on earnings is estimated by decomposing the return to education into a return to the grade completed at entry to service and a return to the post-entry grade increment. Veterans benefits are estimated to increase schooling by roughly 1.4 years and the grade increment is worth roughly 4.3 percent, so that veterans benefits raise annual earnings by approximately 6 percent. This premium appears to accrue primarily to the 77 percent of benefit users who attended college or graduate school.