Many microsatellite loci contain more than a single repeat motif. These compound microsatellites are perhaps most easily explained as arising from simple ancestors. We have uncovered a contrary example in mice of the genus Mus. Sequence analysis of the locus D1MIT29 in most of the members of the genus Mus reveals that this locus is compound in all species except M. musculus, in which it is simple. Moreover, phylogenetic analyses of base substitutions in the non-repetitive flanking region gives trees which are consistent with the previously accepted phylogenetic hypothesis that M. musculus is nested within the subgenus Mus. This confirms that the history of this locus is similar to that of molecular markers previously used in phylogenetic studies of this group and, therefore, demonstrates that the simple state in this lineage is derived from a compound ancestor. We also demonstrate the utility of this type of nuclear sequence variation for phylogeographic studies in the genus Mus. Finally, our sequences reveal homoplasy for size, reemphasizing the danger of using microsatellite size variation alone when the individuals under study are not closely related.