The genetic variability of honeybee populations Apis mellifera ligustica, in continental Italy, and of A. m. sicula, in Sicily, was investigated using nuclear (microsatellite) and mitochondrial markers. Six populations (236 individual bees) and 17 populations (664 colonies) were, respectively, analysed using eight microsatellite loci and DraI restriction fragment length polymorphism (RFLP) of the cytochrome oxidase I (COI)-cytochrome oxidase II (COII) region. Microsatellite loci globally confirmed the southeastern European heritage of both subspecies (evolutionary branch C). However, A. m. ligustica mitochondrial DNA (mtDNA) appeared to be a composite of the two European (M and C) lineages over most of the Italian peninsula, and only mitotypes from the African (A) lineage were found in A. m. sicula samples. This demonstrates a hybrid origin for both subspecies. For A. m. ligustica, the most widely exported subspecies, this hybrid origin has long been obscured by the fact that in the main area of queen production (from which most of the previous ligustica bee samples originated) the M mitochondrial lineage is absent, whereas it is present almost everywhere else in Italy. This presents a new view of the evolutionary history of European honeybees. For instance, the Iberian peninsula was considered as the unique refuge for the M branch during the quaternary ice periods. Our results show that the Apennine peninsula played a similar role. The differential distribution of nuclear and mitochondrial markers observed in Italy seems to be a general feature of introgressed honeybee populations. Presumably, it stems from the social nature of the species in which both genome compartments are differentially affected by the two (individual and colonial) reproduction levels.