This paper examines the relation between chromosomal and nuclear-gene divergence in 28 wild populations of the house mouse semi-species, Mus musculus domesticus, in Western Europe and North Africa. Besides describing the karyotypes of 15 of these populations and comparing them to those of 13 populations for which such information was already known, it reports the results of an electrophoretic survey of proteins encoded by 34 nuclear loci in all 28 populations. Karyotypic variation in this taxon involves only centric (or Robertsonian) fusions which often differ in arm combination and number between chromosomal races. The electrophoretic analysis showed that the amount of genic variation within Robertsonian (Rb) populations was similar to that for all-acrocentric populations, i.e. bearing the standard karyotype. Moreover, divergence between the two types of populations was extremely low. These results imply that centric fusions in mice have not modified either the level or the nature of genic variability. The genetic similarity between Rb and all-acrocentric populations is not attributed to the persistence of gene flow, since multiple fusions cause marked reproductive isolation. Rather, we attribute this extreme similarity to the very recent origin of chromosomal races in Europe. Furthermore, genic diversity measures suggest that geographically separated Rb populations have in situ and independent origins. Thus, Rb translocations are probably not unique events, but originated repeatedly. Two models are presented to explain how the rapid fixation of a series of chromosomal rearrangements can occur in a population without lowering variability in the nuclear genes. The first model assumes that chromosomal mutation rates are between 10(-3) and 10(-4) and that populations underwent a series of transient bottlenecks in which the effective population size did not fall below 35. In the second model, genic variability is restored following severe bottlenecks, through gene flow and recombination.