This paper evaluates the wage effects of the Job Corps program, one of the largest federally-funded job training programs in the United States. The study uses data from the National Job Corps Study, a randomized evaluation funded by the U.S. Department of Labor. The main challenge in assessing the program's impact on wages is sample selection, where wages are only observed for those who are employed, and employment status itself may be affected by the training program. The paper develops a trimming procedure to bound average treatment effects in the presence of sample selection, without requiring exclusion restrictions or a bounded support for the outcome variable. The method identifies the proportion of individuals induced to be selected due to the treatment and then trims the upper and lower tails of the outcome distribution by this proportion. The estimator is shown to be consistent and asymptotically normal, with an intuitive expression for its asymptotic variance. The analysis of the Job Corps program using this method suggests a positive causal effect on wage rates, consistent with the idea that the program raises earnings by increasing human capital rather than solely through encouraging work. The trimming procedure is generally applicable to treatment evaluation problems with non-random sample selection/attrition.This paper evaluates the wage effects of the Job Corps program, one of the largest federally-funded job training programs in the United States. The study uses data from the National Job Corps Study, a randomized evaluation funded by the U.S. Department of Labor. The main challenge in assessing the program's impact on wages is sample selection, where wages are only observed for those who are employed, and employment status itself may be affected by the training program. The paper develops a trimming procedure to bound average treatment effects in the presence of sample selection, without requiring exclusion restrictions or a bounded support for the outcome variable. The method identifies the proportion of individuals induced to be selected due to the treatment and then trims the upper and lower tails of the outcome distribution by this proportion. The estimator is shown to be consistent and asymptotically normal, with an intuitive expression for its asymptotic variance. The analysis of the Job Corps program using this method suggests a positive causal effect on wage rates, consistent with the idea that the program raises earnings by increasing human capital rather than solely through encouraging work. The trimming procedure is generally applicable to treatment evaluation problems with non-random sample selection/attrition.