Children aged 19-42 months were video-recorded while playing in an experimental room equipped with gender-stereotyped toys. Children played first with their mothers and then with a familiar same-or opposite-gender partner. Thirteen categories of social behaviour were coded from the videotape record using a computer-assisted coding system. Twelve categories were combined in an aggregate measure-total social behaviour which increased with age and was more common in same-gender pairs, except among boys in the oldest age group. Subcategory analyses revealed different patterns. Prosocial behaviour was more common in mixed pairs and the frequencies of assertion and withdrawal behaviour provided mirror images. Girls in the first three age groups and boys in the oldest group displayed more ussertion but boys in the first three age groups and girls in the oldest group showed more withdrawal. At all ages boys were more successful in their assertive bids.The preference of young children for same-gender playmates has led to a number of studies which have sought to describe the development of gender-related patterns in various aspects of social interaction. Jacklin & Maccoby (1978) demonstrated that 33-month-olds directed more social behaviour, both positive and negative, to partners of the same gender. However, they found difficulty in explaining how these children identified their partner's gender since the children were strangers and were given no obvious clues such as gender-marked clothes or names.Studying children who attended the same play group Langlois el a / . (1973) found that three-year-old girls and all five-year-olds displayed the pattern described by Jacklin & Maccoby but that three-year-old boys, in contrast, engaged in more social behaviour in mixed pairs than in same-gender pairs. Phinney & Rotheram (1982) also observed children who normally played together and noted that the frequency of social overtures varied both with the gender of the instigator and the target child.The impact of partner's gender on behaviour has been demonstrated at a micro level as well. Wasserman & Stern (1978) have shown that three-and five-year-olds approach more closely, and orient their bodies more frontally, to a peer of the same gender. These studies raise a number of interesting questions relating both to the ages at which gender differences in social behaviour are first observed and to the precise nature of the underlying social processes.In a cross-sectional study we examined the behaviour of children from 19 to 42 months of age, seeking to show the effects of partner's gender on social behaviour at different ages. In order to study established patterns of social behaviour at an early age we recruited children who knew each other and were already accustomed to playing together. & Lucas, 1975;Hartup, 1978). Our design allowed us to investigate the impact of age on total amounts of social behaviour which we expected would increase over the age span studied (19-42 months). We also varied pair composition and observed boy-boy, g...