The long-term life outcomes of disadvantaged youth lag far behind their higher-income peers. Providing low-income youth with jobs, especially over the summer, has long been a popular policy tool to address socio-economic disparities. There are a number of reasons to think summer jobs might improve outcomes; they provide income, information about returns to schooling, connections to employer networks, and training in work and social skills; and they keep youth busy when they might otherwise be idle. Yet there is almost no rigorous evidence on the effects of summer jobs on youth outcomes, especially on criminal behavior. This paper provides first-of-its-kind causal evidence on the effects of a summer job on disadvantaged adolescents. I randomly assign 1,634 applicants from 13 low-income, high-crime high schools in Chicago to be offered a part-time subsidized summer job, a job plus cognitive behavioral therapy-based programming, or the status quo (no additional services). Results are mixed, with a 3 percentage-point (30 percent) decrease in summer school enrollment for those assigned to treatment, but also a huge proportional drop in violent-crime arrests after 7 months (2.7 fewer arrests per 100 youth, a 49 percent decrease). A full benefit-cost analysis awaits longer-term follow-up data, though it seems plausible that if these results persist, the fall in such a socially-costly outcome (violent crime) could generate benefits that outweigh program costs.