This paper illustrates the process of program evaluation for ProJoven, the Peruvian youth labor training program. The program provides beneficiaries with basic three-month training in low-skill occupations and with internship opportunities. ProJoven's design promotes gender equality by encouraging female participation in training for traditionally male-dominated occupations and by providing subsidies so mothers with children can participate. Complementing detailed fieldwork in search of the appropriate control group, the econometric work implements a two-stage matching procedure that includes propensity scores (on the first stage), and gender and labor income (on the second one). The matching on gender allows identification of differentiated program impacts on men and women. The matching on income attacks the problem of Ashenfelter's Dips. The evaluation shows substantial differences in ProJoven's impact for men and women. Eighteen months after participation in the program, employment rates for women improved by about 15% (while employment for men reduced by 11%), gender occupational segregation reduced by 30% and women's labor income improved by 93% (while men's earnings increased by 11%). On the other hand, the cost of the promotion of gender equality represented only 1.5% of ProJoven's total budget. These results suggest that labor training programs that promote equal gender participation have disproportionately positive effects on outcomes for women trainees in a labor market with substantial gender differences.