The ecological interactions parasitic insects have with their hosts may contribute to their prodigious diversity, which is unrivaled among animals. Many insects assumed to be polyphagous generalists have been shown to consist of several differentiated races, each occupying a different host-niche. The sunflower maggot fly, Strauzia longipennis, has long been thought to consist of two or more races due to its substantial intraspecific morphological variation. Here, we use nuclear and mitochondrial markers to test the hypothesis that S. longipennis is a complex of two or more partially reproductively isolated races. We collected S. longipennis flies as pupae from roots of Jerusalem artichoke (Helianthus tuberosus) and as adults swept from leaves of mature H. tuberosus across the breadth of a field season. Flies were scored for morphological variety (typica or vittigera), mitochondrial haplotype (A or B) and a panel of 176 AFLP loci. Bayesian clustering and neighbor-joining phylogenetic analyses of AFLP data supported the existence of at least three, possibly four, genetic races of Strauzia (clusters I, II, III, and V), as well as a small number of putative interracial hybrids (cluster IV). Clusters I and III each consisted of flies of both morphological varieties and both haplotype groups, while flies in cluster II were all of variety typica and all but one was of mitochondrial haplotype B. Flies in cluster II were also collected only as adults on H. tuberosus and not among flies reared from pupae collected from H. tuberosus roots, suggesting that they use a different plant as their larval host. Mean capture date was significantly different between flies of each genetic race, indicating that partial allochronic isolation may be one contemporary barrier to gene flow between races. Evidence that mitochondrial genomes and morphological traits have moved between lineages implies a model of speciation-with-gene-flow for S. longipennis races.