We study the effect of a job training program for low income youth in Cordoba, Argentina. The program included life-skills and vocational training, as well as internships with private sector employers. Participants were allocated by means of a public lottery. We rely on administrative data on formal employment, employment spells and earnings, to establish the effects of the program in the short term (18 months), but also – exceptionally for programs of this type in Latin America and in developing countries in general – in the medium term (33 months) and in the long term (48 months). The results indicate sizable gains of about 8 percentage points in formal employment in the short term (about 32% higher than the control group), although these effects dissipate in the medium and in the long term. Contrary to previous results for similar programs in the region, the effects are substantially larger for men, although they also seem to fade in the long run. Program participants also exhibit earnings about 40% higher than those in the control group, and an analysis of bounds indicates that these gains result from both higher employment levels and higher wages. The detailed administrative records also allow us to shed some light on the possible mechanisms underlying these effects. A dynamic analysis of employment transitions indicates that the program operated through an increase in the persistence of employment rather than from more frequent entries into employment. The earnings effect and the higher persistence of employment suggest that the program was successful in increasing the human capital of participants, although the transient nature of these results may also reflect better matches from a program-induced increase in informal contacts or formal intermediation.