The present study aims to provide a systematic understanding of how perspective-taking ability contributes to primary-school students' cooperative behaviours and learning outcomes. The present study is frontline as we combined person-oriented (e.g., describing patterns of behaviours based on individual characteristics), process-oriented (e.g., examining factors affecting the quality of cooperative behaviours), and effect-oriented (e.g., examining the effect of cooperative learning on individual learning outcomes) analytical approaches within one research framework. In addition, we adhered to the multi-dimensional nature of perspective-taking ability and differentiated between social and cognitive perspective-taking ability while taking into account the contribution of perspective-taking ability at both the individual level and group level (i.e., heterogeneous and homogeneous perspective-taking ability groups) to cooperative behaviour profiles and learning outcomes of primary-school children. Based on transcribed episodes of interaction of 115 fifth-grade students, four different profiles of cooperative behaviours were discerned: captains, hard workers, switchers, and passive participants. We found that these profiles are related to perspective taking conceptualized at the group level, but not to individual-level perspective-taking ability. Profile membership, cognitive perspective-taking ability, and group-level perspective-taking ability could not predict students' learning outcomes. Social perspective-taking ability and reading comprehension did positively predict learning outcomes. Our findings add to existing knowledge as they suggest that the influence of perspective-taking ability on cooperative behaviours and learning outcomes is susceptible to the conceptualization (i.e., cognitive vs. social) and measurement level (i.e., individual vs. group level) of perspective-taking ability.