76 randomised experiments conducted in developing-country primary schools from the mid-1970s to 2013 were identified and coded. The experiments evaluated the impact of 110 school-based treatments on language and mathematics test scores, as compared with “business as- usual” in the same settings. The treatments included instructional interventions, health interventions, and incentive-based interventions. On average, monetary grants and deworming had effects that were close to zero and statistically insignificant. Nutritional treatments, treatments that provided information to parents or students, and treatments that improved school management and supervision had small mean effect sizes (0.04-0.06) that were not always robust to controls for study moderators. The largest mean effect sizes included treatments with instructional materials (0.08); computers or instructional technology (0.15); teacher training (0.12); smaller classes, smaller learning groups within classes, or ability grouping (0.12); student and teacher performance incentives (0.10); and contract or volunteer teachers (0.10). Meta-regressions that controlled for treatment heterogeneity and other moderators suggested that the effects of materials and contract teachers, in particular, were partly accounted for by composite treatments that also included teacher training and class size reduction. A caveat is that interventions like deworming and school lunches often affected enrolment and attainment independently of learning, implying that student time is not always used productively in schools. There is insufficient data to gauge the relative cost-effectiveness of categories of interventions.