Household income has been shown to matter for children's school enrolment, in particular in settings where households face tight liquidity constraints caused by the lack of insurance and limited possibilities to smooth consumption through credit and savings. However, so far only few studies have made an effort to quantify the income elasticity of school enrolment, in particular in the Sub-Saharan African context. The empirical problem in identifying the causal impact of income on enrolment is to control for parental ability, which is largely unobserved, and to deal with reverse causality and measurement error. This paper uses for identification a natural experiment in Burkina Faso, a country with particularly low enrolment rates. The results show that naive estimates largely underestimate the true income elasticity of school enrolment. The results can provide a basis for safety net policies. (Contains 6 tables and 2 figures.)