The question about the rationality of crime and violence is not only a controversial issue but also has strong policy implications. Despite the large number of rationality studies in criminology, there is little empirical evidence on how rationality can be moderated by non-rational mechanisms. Additionally, many studies examining the influence of rational choice on crime suffer from several methodological problems associated with small and biased samples, limitations of the dependent variable, varying operationaliations of rationality, and scarce inclusion of validated non-rational causal mechanisms. Finally, there is a lack of cross-cultural validity of rational and non-rational predictors of crime since most studies have been conducted in high-income societies. Little research has examined how well these explanations can fit the socio-economic, cultural and institutional characteristics of the Latin-American context. The goal of this study was to examine a rational choice model of crime and its interactions with three well known non-rational causal mechanisms in criminology: morality, legitimacy and self-control. The study involved the application of a survey on 2,204 9th grade youths from a representative sample of high schools in a middle income society in Latin America: Montevideo, Uruguay. The questionnaire was an adaptation of the Zurich Project on the Social Development of Children Study. The analysis was conducted using count regression models including principal and interaction effects based on a hierarchical or blockwise entry method. Results indicated that rationality plays a robust but modest explanatory role even after including socio-demographic variables and the three non-rational predictors. Rational choice theory was supported as a general theory that accounts not only for general crime but also for property and violent crimes. Additionally, rationality had stronger effects than two of the other three non-rational mechanisms: legitimacy and morality. Different dimensions of rationality were examined. Analyses showed that inner costs and peer reactions have significant associations with all types of youth crime, whereas formidability, parents’ reactions and police reactions did not. Finally, the analysis of interactions suggested that the link between rationality and youth crime is mostly unaffected by self-control as a moderator, and moderately conditioned by legitimacy and morality, particularly the latter. Findings, although provisional in Latin-American context, may provide new insights for future research in rational and non-rational mechanisms of youth crime. Research and policy implications of these findings are discussed.