One rationale for work-focused welfare reform was human capital theory: work today should raise experience tomorrow, which should raise future wage offers and reduce welfare dependency. Yet few studies have estimated the effect of welfare reform on wages. I approach the problem using a novel sample selection estimator based on reservation wage data. Reservation wages solve the selection problem using bivariate censored regression methods without the need for exclusion restrictions. Whereas OLS and conventional sample selection estimates suggest that reform had little effect on wages, the reservation-wage-adjusted estimates suggest that Florida's welfare reform experiment raised wages by about 4%.