Data from a natural experiment are used to demonstrate how a sudden change in education policy in Burkina Faso is useful in estimating the effect of maternal education on child health. Indeed, a major problem in estimating the effect of maternal education on child health is that unobserved factors may affect maternal education and child health simultaneously, causing endogeneity bias. Most studies on the relationship between maternal education and child health have used instrumental variables to address the endogeneity problem. Because it is very difficult to find instrumental variables that clearly satisfy the requirements of “correlation with maternal education” and “non- correlation to unobserved factors”, instrumental variables do not guarantee a solution to the endogeneity problem. A way around this is the use of natural experiment as an identification strategy. The results show that mother’s education significantly and positively affects child weightfor-height (WHZ, an indicator of current malnutrition). The effect of mother’s education on child height-for-age (HAZ, an indicator of chronic malnutrition) is positive but statistically insignificant, suggesting no direct effect of mother’s education on child’s HAZ. Per capita household expenditures seem to be the pathway through which mother’s education affects children’s HAZ. For WHZ, the mother’s education variable was found to be significant even after including all pathways in the regression, indicating that mother’s education, by itself, has a strong causal effect on child health or there are other pathways that the data used did not allow us to investigate. The threshold effects estimation indicate the largest impacts of mother’s years of education at 13 years of education for WHZ and 12 years for HAZ. Thus, after taking costs into consideration, education policies targeted to girls should focus on trying to maintain them in school longer than the six years that has been the standard so far.