Despite the research exploring the benefits of using cooperative learning in secondary physical education, several research gaps still remain, namely, limited research on its effects on prosocial behaviours, the impact on adolescents as they grow up, and the lack of assessment of differently structured cooperative learning frameworks. The goal of this study was to compare high- versus low-structured cooperative learning frameworks and assess their impact on adolescents' prosocial behaviours at different ages. The study followed a quasi-experimental pre-test--post-test design. Two hundred and eighty-six students (150 girls) participated and were distributed into four groups: Control 1 (year-8 students), Control 2 (year-10 students), Experimental 1 (year-9 students) and Experimental 2 (year-11 students). All groups experienced the same Acrosport unit, but the control groups were within a low-structured cooperative learning context and the experimental groups were within a high-structured cooperative learning framework. Cooperative learning and prosocial behaviours were measured before and after the intervention. Results showed that students who experienced a high-structured framework at a younger age significantly increased their scores on the five variables that mediate the effectiveness of cooperative learning, and their prosocial behaviours. The older the adolescents, the fewer the changes in cooperative learning, and with no changes in prosocial behaviours. Students who experienced a low-structured framework did not improve their cooperative learning and decreased their empathy, social relations, and leadership at younger ages. In conclusion, cooperative learning must be properly structured to produce a positive impact using heterogeneous working groups, teachers' feedback and shared regulation. However, older adolescents need specifically designed cooperative learning contexts with longer learning units.