This research looks at the potential of peer interaction practices in improving narrative competence by analyzing the efficacy of peer learning on children's oral narrative productions. Gains on a macro-level (structure and coherence of the narrative) and a micro-level (cohesion of the narrative) were analyzed. Fifty-six primary school children participated in this study. Each child told a narrative either individually (individual condition) or while interacting with a peer (joint condition). We explored whether children produced longer, more structured, coherent and cohesive narratives in a joint condition rather than individually, and in which condition the joint task was more beneficial for children's narrative competence in terms of narrative scores in the individual condition, discrepancy between the members of the same pair, and quality of the interaction. The advantage of peer learning does not derive from the direct comparison of the individual versus the joint condition but depends on specific conditions: the joint condition was beneficial for individuals with lower individual competence and for pairs with a high discrepancy between individual scores. Children's quality of interaction did not seem to influence the efficacy of peer learning on their narrative competence.