In Americas imprisonment policy has increasingly been adopted to deal with law offenders, and a steep increase in countries incarceration rates has been observed since the 1980s. Nevertheless, we still lack evidence on how such penalty affects labour market outcomes of ex-inmates, specially in developing countries. This work exploits the variation in detention tendencies of randomly-assigned judges as an instrumental variable to estimate the impacts of incarceration on formal labour market outcomes.We construct a unique panel dataset merging data from São Paulo city court, containing judicial sentence information with formal employment records. Essentially, we compare individuals who were sentenced to prison with those who were not. We find that the immediate negative impact of sentence to prison on employment tends to fade over time and disappear after release.