This paper estimates the relative effectiveness of private and public primary schools in Kenya using data from 4,433 Grade 6 schoolchildren. Using ordinary least squares as a baseline model, we use the proportion of private schools in a district as an instrument in a Heckman two-stage correction framework, as well as propensity score matching models to correct for selection bias. There is a positive private school effect across all models. In the corrected models, we find that private school pupils outperform their public school counterparts by between 0.24 and 0.52 standard deviations.