Locating birthplaces using genetic parentage determination can increase the precision and accuracy with which animal dispersal patterns are established. We re-analyse patterns of movement away from the birthplace as a function of time, sex and population density for a sample of 303 banner-tailed kangaroo rats, Dipodomys spectabilis. We located birth sites using a combination of likelihood-based parentage analysis with live-trapping of mothers during the breeding season. The results demonstrate that natal-breeding site distances are density dependent in this species; in particular, both sexes emigrate earlier in the year, and females disperse farther than males, at low population densities. Banner-tailed kangaroo rats were chosen as a study system because live-trapping easily detects maternal and offspring locations; nevertheless, parentage analysis reveals that some offspring evade early detection and move substantial distances before their first capture. In a few cases, the approach even detects dispersal out of the natal ‘deme’ prior to first capture. Parentage analysis confirms the extreme philopatry of both sexes but indicates that prior estimates of median dispersal distance were too low. For D. spectabilis, more accurate location of individual birthplaces clarifies patterns of sex bias and density dependence in dispersal, and may resolve apparent discrepancies between direct and indirect estimates of dispersal distance. For species in which mothers can be more reliably trapped than juveniles, using offspring genotypes to locate parents is a novel way that genetic techniques can contribute to the analysis of animal dispersal.