We present evidence on how The Quantum Opportunity Program (QOP hereafter) worked in the US. While the program was regarded as successful in the short-term, in the long-run its educational results were modest and its effects on risky behaviors detrimental. Exploiting control group's self-reported drug use while in school, we evaluate whether the program worked best among those with high-predicted risk of problem behavior. We find QOP to be extremely successful among high-risk youths as it managed to curb their risky behaviors during high-school and, by doing so, it persistently improved high-school graduation by 20 percent and college enrollment by 28 percent. In contrast, QOP was unsuccessful among youths in the bottom-half of the risk distribution as it increased their engagement in risky behaviors while in high-school. Negative peer effects are possibly an explanation behind these results. Finally, negative peer effects also seem to explain the longer-run detrimental effects of QOP on risky behaviors.