This paper investigates the impact of working while in school on learning outcomes through the use of a unique micro panel dataset of students in the São Paulo municipal school system. The potential endogeneity of working decisions and learning outcomes is addressed through the use of a difference-in-difference estimator and it is shown that the results are robust. A negative and significant effect of working on learning outcomes in both math and Portuguese is found. The effects of child work from the benchmark regressions range from 3% to 8% of a standard deviation decline in test score which represents a loss of about a quarter to a half of a year of learning on average. Additionally, it is found that this effect is likely due to the interference of work with the time kids can devote to school and school work.