The objective of this paper is to assess the impact of working in the twelve months preceding the date of leaving high school, either as a graduate or as a dropout, on the probability of graduation. To do so, I use Statistics Canada's 1991 School Leavers Survey and its 1995 Follow-up. Given that both the decision to graduate and the decision to work are endogenous variables, I use local labor market conditions as an exclusion restriction to study the sensitivity of the results to different estimation techniques in a system of endogenous limited-dependent/qualitative variables. While all estimation methods lead to roughly the same qualitative conclusion, relaxing some of the underlying distributional assumptions in favor of semi-parametric (or less restrictive) methods generally leads to larger impacts than what I get with full maximum likelihood techniques. In conclusion, while previous work using U.S. data points towards somewhat ambiguous effects, the results here show a strong negative effect of working while in school on the probability of graduation, especially for men. This negative effect shows up both when I use a dummy for work activity while in school or when I use hours worked directly.