Lowering dropout levels would not necessarily make society better off and, consequently, student aid policy should not be evaluated by its effects on dropout. Analysis of the model of postsecondary enrollment and completion developed in this paper results in two striking findings. First, making reduction of dropouts the policy goal yields the perverse conclusion that it would be best to eliminate student aid entirely because eliminating aid would make enrollment less attractive than work and hence reduce the number of students who choose to enroll. The students who choose to work rather than enroll are those with the lowest completion probabilities, so eliminating the aid shifts the composition of those enrolled toward those students with the highest completion probabilities. Second, a policy that reduces dropout probabilities does indeed lower dropout among those students who would have enrolled in the prepolicy regime; at the same time, however, it also induces new students to enroll. The dropout level rises if the number of induced enrollees who drop out exceeds the gain in completion among existing enrollees. The dropout rate rises if the completion probabilities of induced enrollees are sufficiently lower than those of existing enrollees. Completion subsidies may induce students to choose programs with high pass rates and thus may inhibit good matches between students and careers. The document develops a model elaborating on the idea that completion of schooling is exogenous, applies the model to the study of enrollment and completion, and extends the analysis to allow for the possibility that completion is partly endogenous. An appendix provides proofs for the two major propositions. Seven references are included. (CML)