This paper presents estimates of production functions of reading and mathematics test scores to assess the effects of supply-side interventions, such as the provision of a community-based school management programme, bilingual education and multigrade teaching, on student learning in Guatemala. The efficiency and consistency of the estimates is improved by using seemingly unrelated regressions to account for potential correlation between test scores. Results indicate that the community-based management programme, bilingual schools and traditional schools produce comparable learning of first-graders. However, in the case of third-graders the community-based school management programme and bilingual schools produce less learning than traditional schools. The learning of both first- and third-graders is found to be diminished in the multigrade teaching approach.