This thesis aims to make a contribution to our current understanding regarding leisure-time pedagogues’ work with children’s relationships. Here focus is placed on the collective action repertoire as expressed by two different groups of leisure-time pedagogues. The research questions addressed are:   How do leisure-time pedagogues work with children’s relationships and how do they view such relationships? What communities do they identify? What qualities in children’s relationships do they regard as desirable?  The theoretical framework of this study is based on Wenger’s (1998) social theory regarding learning in communities of practice, as well as Gergen’s theory of relationships as an inevitable part of human existence. Consequently, within this thesis relationships are viewed as mainly negotiated and are therefore dynamic in character. The methodological approach is ethnographic, focusing on leisure-time pedagogues’ work with and talk about children’s relationships. The study is based on data derived from eight leisure-time pedagogues divided into two different work teams of four pedagogues, covering two different leisure-time centres and involving 60 children of 6-11 years of age.   Final analysis is based on concepts derived from both Wenger’s and Gergen’s theories, for example; action repertoire, relational qualities, communities of practice and shared interests. Results suggest that the communities of practice and the alliances that pedagogues identify are often gender-related and built on common interests. The pedagogues’ action repertoire illustrates a desire to facilitate encounters between children and promote harmonious relationships. Leisure-time pedagogues support relationships characterized by consensus, respect, confidence and adaptation to rules, whereas those marked by conflict, disharmony, breaking rules or aggressions are counteracted. Various notions emerge in pedagogues’ action repertoire based on normative thinking, where different relational competences are ascribed to children. Some children are described as relationally competent while other children, who do not adapt themselves in a desirable manner, are described as having difficulties adjusting their relational competence. The results also reveal a lack of guidelines for handling the variety of differences in children’s relationships and display the lack of a common professional language for verbalizing children’s relational work.