In a population of the monogynous, polyandrous ant Cataglyphis cursor, we analysed the spatial genetic structure of queens, colony fathers and workers at a microgeographical scale to infer the extent of sex-biased dispersal and to assess the impact of limited dispersal on the patterns of relatedness within the colony. To this end, four microsatellite markers were scored for the queen and an average of 26 workers from each of 35 mapped colonies. We used pair-wise kinship coefficients between all pairs of genotypes, including the reconstructed colony father genotypes (1) to test and quantify isolation by distance patterns within each sex or caste through the analysis of kinship–distance curves, and (2) to compute the average relatedness between categories of colony members. The kinship–distance curve was much steeper for colony queens than colony fathers, indicating male-biased dispersal. However, colony fathers also displayed a non-random spatial genetic structure, so that even males show some dispersal limitation at the scale of the population, which extends over less than 250 m. The degree of relatedness between the different sexes and castes of colonies was well predicted from the number of mates per queen and the inbreeding of queens, and the impact of limited dispersal was very weak at this scale of observation. We discuss the interest of kinship–distance curves to assess sex-biased dispersal on a local scale and we compare our results with large-scale analyses of genetic structure in Cataglyphis cursor and other monogynous ant species. © 2008 The Linnean Society of London, Biological Journal of the Linnean Society, 2008, 93, 465–473.