The sexual segregation recorded outside the rut in many gregarious ungulates is most often considered as ecologically determined, i.e. resulting from sexual differences in habitat use that enable males and females to improve their fitness in relation to their respective vulnerability to predators and their energy requirements. However, sexual segregation might also result from alternative causes, and especially from spatial and social mechanisms. Accordingly, we analysed the distribution and aggregation pattern of a fallow deer population living in an enclosed heterogeneous forest. Nested stochastic models were used to quantify the effects of habitat preferences, within-patch space use and attraction/repulsion between animals. The analysis first shows that spatial distribution of males and females was not only the outcome of sex-related habitat preferences: a purely spatial process, possibly involving space attachment, was clearly also at work. Furthermore, an attraction between animals, and especially between individuals of the same sex, appears as the primary cause of the observed group sizes and compositions. This social process partly explains, in addition, the pattern of space use observed for each sex in the studied population. So, we propose to reverse the common assumption according to which animal grouping is a simple consequence of habitat exploitation, into the assumption that a sex-dependent attraction between individuals and a process of space attachment generate not only the grouping pattern but also partly the habitat use observed in various ungulate species