We examine if compulsory schooling laws (CSL) necessarily lower crimes. We focus on violent youth crime (homicides by assault and guns) among 15-19 years age group in all Brazilian municipalities over 2000-13, taking advantage of the 2009 Brazilian Constitutional Amendment that required introduction of compulsory high schooling of 15-17-year-olds by 2016. Only about 53% municipalities adopted the Amendment by 2013. Difference-in-difference estimates with municipality fixed effects to account for the endogenous adoption of the Amendment by municipalities show small treatment effects for homicides, but insignificant effects for homicide rates in the full sample. In the absence of any significant increase in income/employment among this age group, we attribute this to the incapacitation effect of CSL, which was, however, weakened by overcrowding in day and night schools in treated municipalities after 2009. In contrast, poorer treated municipalities witnessed increased class size, worse school performance and increased crime too. The crime reduction effects of CSL thus crucially depend on whether/how it affects class size and school quality especially in less promising jurisdictions.