In this paper, our main objective is to test the hypothesis that child labor can be a rational response to low returns to formal education in Mali. To this end, after a literature review, we build a flexible conceptual model that explicitly links the child labor supply to the comparison of the expected returns to education with child labor. The empirical analyses are performed using the Malian permanent household survey. Estimations suggest that when returns to education are high in local labor markets, or when parents earn more than expected relative to their education level, the probability of child labor falls.