In 1976 Congress terminated the largest federal program of direct student assistance, the GI Bill. This study followed in the wake of growing dissatisfaction with the existing program and of recruiting difficulties experienced by the services during the late 1970s. This report presents the history and results of the Educational Assistance Test Program, a large-scale experiment conducted by the Department of Defense during fiscal year 1981. The experiment was mandated by Congress to determine the effectiveness of expanded postservice educational assistance programs for attracting more high aptitude high school graduates into military service. The results suggest that a narrowly targeted program can increase enlistments by an amount that is out of proportion to the limited number of jobs covered, and do so at a cost per additional recruit that probably is lower than that for a general program under which everyone brought in by the program may ultimately receive the benefit. Implications for future policy are discussed which provide a framework for the design of a relatively cost-effective program of military educational benefits. (PN)