Researchers focused upon the work-dropping out connection tend to show a U-shaped relationship between the likelihood of dropping out and the number of hours worked outside school, with a higher exit rate for both non-working students and for students whose working hours pass a critical threshold. Yet the data typically used by these researchers are drawn mainly from cross-sectional surveys, and as a result does not allow for any causal interpretation. The present article uses an event history analysis of Canadian longitudinal data covering seven years of a cohort, and offers original findings on the causal work-dropping out relationship at the university level. We find evidence showing that the evolution of the exit rates and the factors influencing the decision to quit a particular university programme differ substantially between students who want to enrol in another programme and those who do not. For the latter, we observe a critical threshold of 24h of work, beyond which negative effects in terms of non-completion start to appear. We find no negative effects arising from not working vs. working a few hours. Our findings thus tend to show that the higher exit rate among non-working students evidenced in cross-sectional data should be attributed to the fact that academic difficulties cause some potential university dropouts to stop working and to devote more time to school.