This paper studies the linkages between schools' inputs and students' performance in Latin America. We exploit the richness of PISA 2012 questionnaires at the student and school level to study the association between a different set of inputs and students' performance in mathematics. First, this research shows that students' characteristics and their environment (i.e. sex, age and economic, social and cultural status of students and schools) explain close to 30% of the variation in education performance in Latin America, a higher percentage than in OECD and other economies which participated in PISA 2012. Second, after controlling for students' characteristics and their environment, our results show that in Latin America, some non-traditional school inputs, such as the feedback provided by the principal to the teacher, weekly instructional time or the attitude and motivation of teachers, are associated with student performance, whereas more traditional inputs (e.g. school infrastructure, share of certified teachers and teacher qualifications) are not always related to better learning outcomes. These findings suggest that some pedagogical initiatives, which are also more cost-effective, could improve students' performance in the region.