In this paper, I analyze the effect of teamwork as a teaching practice on the students performance. A few previous studies using field experiments have shown that cooperation may improve students’ grades, but in small scale field experiments. I use a large sample for 17 countries from PISA, exploiting that detailed information about teaching practices is reported through a teacher questionnaire for the first time in the 2015 wave. My estimates control for students fixed effects in order to account for the fact that teachers can adapt their methods of teaching depending on the ability their students have. I find a positive and significant effect of working in groups in the students scores. Additionally, I measure the impact of different ways to carry out the grouping of students by the professor. I find that no teacher feeback («appreciations») have a negative impact on scores. Finally, results does not seem to point out to an «optimal» way of grouping students to carry out the teamwork: grouping according to specialization of each member or because they have different/similar abilities does not have a significantly different impact on students’ outcome.