The hybrid zone between the two subspecies of mice Mus musculus domesticus and Mus musculus musculus, which has been studied extensively in Denmark, crosses Europe to the Black Sea through the Alps and the Balkans. Two hundred and seventy-nine animals were captured in 22 localities along a transect across the Balkans. The animals were characterized for seven diagnostic nuclear loci by protein electrophoresis and by restriction pattern analysis of their mitochondrial DNA. The nuclear data show a sharp transition between the two subspecies, most of the variations in allele frequencies (from 0.9 to 0.1) occurring within a 36-km section of the transect. The introgression varies from one locus to the other and is more pronounced, in terms of distance, in M. m. musculus territory. Mitochondrial DNA introgression is important but occurs in one direction only, i.e. from M. m. musculus to M. m. domesticus, while a cytoplasmic transfer from M. m. domesticus to M. m. musculus has been reported. A previous study showed that no Y chromosome introgression occurs. The different behaviour of these three types of markers could be due to the interaction between selection against hybrid genomes and meiotic recombination. Objectively, it would appear that the genes that can introgress are neutral or nearly so and have been separated from deleterious genes they were linked to by recombination. This could explain the differential introgression between autosomal loci. The mitochondrial and Y chromosomes undergo no or very little recombination and each is transmitted as a whole. Their degree of introgression is thus indicative of the intensity of selection resulting from the amount of functional differentiation between the two taxa, which seems to be strong for the Y chromosome and weak for mitochondrial DNA. We propose that the asymmetry of nuclear introgression is due to different population structures. As M. m. musculus is relatively less structured, the rapid spreading of introgressed genes would be favoured. Such a scheme, however, can hardly account for the unidirectionality of the mitochondrial flow, which could be due to sex-dependent behaviour.