Bat-swarming sites where thousands of individuals meet in late summer were recently proposed as 'hot spots' for gene flow among populations. If, due to female philopatry, nursery colonies are genetically differentiated, and if males and females of different colonies meet at swarming sites, then we would expect lower differentiation of maternally inherited genetic markers among swarming sites and higher genetic diversity within. To test these predictions, we compared genetic variance from three swarming sites to 14 nursery colonies. We analysed biparentally (five nuclear and one sex-linked microsatellite loci) and maternally (mitochondrial D-loop, 550 bp) inherited molecular markers. Three mtDNA D-loop haplolineages that were strictly separated at nursery colonies were mixed at swarming sites. As predicted by the 'extra colony-mating hypothesis', genetic variance among swarming sites (V ST ) for the D-loop drastically decreased compared to the nursery population genetic variance (V PT ) (31 and 60%, respectively), and genetic diversity increased at swarming sites. Relatedness was significant at nursery colonies but not at swarming sites, and colony relatedness of juveniles to females was positive but not so to males. This suggests a breakdown of colony borders at swarming sites. Although there is behavioural and physiological evidence for sexual interaction at swarming sites, this does not explain why mating continues throughout the winter. We therefore propose that autumn roaming bats meet at swarming sites across colonies to start mating and, in addition, to renew information about suitable hibernacula.