We analyze whether field of study assigned at age 16 impacts mental health in adulthood. Using a regression discontinuity design that exploits GPA cut-offs, we find that admission to the preferred study field improves mental health, lowering both the incidence of antidepressant prescriptions and of mental health-related hospitalizations. Engineering contributes strongly but not uniquely to the positive results. As for mechanisms, earnings explain 40% of the estimates, but earlier proposed hypotheses based on school-age peer characteristics have little explanatory power. Our findings imply that restrictions on individuals' choices, to improve human capital allocations, entail costs that may have been underestimated.