The Rhagoletis pomonella species group has for decades been a focal point for debate over the possibility of sympatric speciation via host shift. Here I present the first extensive analysis of genetic (allozyme) divergence in the pomonella group, including all known taxa/populations except the allopatric Mexican population of R. pomonella. The phylogeny is estimated for all four described species (pomonella, mendax, zephyria, and cornivora) plus two undescribed species (the "flowering dogwood fly" and "sparkleberry fly"). Allozyme data for two additional populations of uncertain status (the "plum fly" and "mayhaw fly") are presented for the first time. Two data sets were analyzed, one for 17 loci from 77 populations and one for an additional 12 loci for a subset of 12 of these populations, with more than 4000 flies analyzed in total. Interspecific Nei unbiased genetic distances were generally small, being as low as 0.040. No fixed autapomorphic alleles beyond those already known for R. cornivora and R. zephyria were revealed in the new data, but several loci displaying frequency patterns useful in discriminating the species were discovered. The phylogenetic placement of the flowering dogwood fly differed depending on whether a molecular clock was assumed (UPGMA of Nei distance) or not assumed (frequency parsimony) for tree building. Other than this, however, trees under either assumption were essentially identical. The best tree was used to test the prediction of the sympatric speciation hypothesis that sister taxa should be broadly sympatric. This prediction was not rejected, but the best tree was weakly supported by bootstrap analysis. An unexpected finding was that R. pomonella populations representing ends of its strong latitudinal clines did not cluster together. One possible explanation is that the current R. pomonella is the result of a genetic fusion of two previously isolated, genetically differentiated populations. Such a fusion prior to the origin of the other species in the group could contribute to the poor resolution of the phylogeny.