In this paper we evaluate the impact of grade span configuration on studentoutcomes. We build a model which shows that the more homogeneous the cohort ofstudents the higher the share of them which achieves the minimum level of academicperformance required in an exam. We then test this theoretical finding by comparingperformances of 5th-grade students of elementary schools (1st to 5th grade) and thoseof elementary-middle schools (1st to 9th grade) in Brazil. PSM and PSM with Diffin-Diff methodologies are used to control possible biases. We find that elementaryschools present better results in Portuguese language and mathematics standardizedscore tests, higher passing rates and lower dropout rates than elementary-middleschools, which corroborates the theoretical results. The robustness of estimatesis checked through several tests and alternative specifications. Finally, we foundevidence that the main mechanisms behind our results are related to alternativepedagogical practices and school management policies.