Many studies have explored the links between music and children's outcomes; however, study designs have not been sufficiently rigorous to support causal findings. This study aims to assess the effects of a large-scale music program on children's developmental functioning in the context of high rates of exposure to violence. The paper describes the results of an experimental evaluation of Venezuela's National System of Youth and Children's Orchestras. The curriculum of the program, better known as "El Sistema," emphasizes social interactions through group instruction and group performances. The randomized control trial was conducted in 16 music centers between May 2012 and November 2013. In total, 2914 children ages 6-14 participated in the experiment, with approximately half receiving an offer of admission to the program in September 2012 and half in September 2013. The treatment group children participated for one semester more than the control group children. After 1 year, full-sample ITT estimates indicate improved self-control (by 0.10 standard deviations) and reduced behavioral difficulties (by 0.08 standard deviations), both significant at 10% after controlling for multiple hypothesis testing. There were no full-sample effects on other domains. Sub-sample effects are larger among (1) children with less-educated mothers and (2) boys, especially those exposed to violence at baseline. In the latter subgroup, we find lower levels of aggressive behavior. We find that the program improved self-control and reduced behavioral difficulties, with the effects concentrated among subgroups of vulnerable children. The results suggest the importance of devising mechanisms to target resources to the most vulnerable children.