We study the causal impact of network centrality on educational outcomes using field experiments in primary schools in Bangladesh. After obtaining information on friendship networks, we randomly allocate students into groups and give them individual and group assignments. We find that groups that perform best are those whose members have high Katz-Bonacich and key-player centralities. Leaders are mostly responsible for this effect, while bad apples have little influence. Own Katz-Bonacich centrality is associated with better individual performance only if it is above the average centrality of the group. Further experiments reveal that leadership, as measured by network centrality, mainly captures non-cognitive skills, especially patience and competitiveness.