This paper uses household, school, and test score data from Bangladeshto compare and contrast the effectiveness of NGO-run and state-run schoolsin the provision of primary education. I study how the entry of NGOs inprimary education has affected educational outcomes of girls and examine themechanisms which account for the relative performance of NGO versus stateschools in improving female educational outcomes. The results show that theentry of NGO schools has significantly increased girls' enrollment as comparedto boys. Constructing cohorts from cross-sectional data using year of birth andyear of NGO school establishment, I show that cohorts which were exposed toNGO schools have higher probability of enrollment and the effect operatesmainly through girls. The two most prominent characteristics of NGO schoolsthat encourage girls' enrollment are the high percentage of female teachers andhaving Parent-Teacher Associations (PTAs). NGO schools show strong effectsin improving children's test scores.