Herbivorous insects are abundant and diverse and insect-host plant associations tend to be specialized and evolutionarily conserved. Some authors suggested that generalist insect lineages tend to become specialists, with host specialization leading to an evolutionary dead-end for the parasite species. In this paper, we have examined this tendency using a phylogenetic tree of Tomoplagia (Diptera: Tephritidae), a parasite of asteracean plants. We have tested the trend towards specialization in different hierarchical degrees of host specialization. The topology of the tree, the inference of ancestral hosts, and the lack of directional evolution indicated that specialization does not correspond to a phylogenetic dead-end. Although most Tomoplagia species are restricted to a single host genus, specialization does not seem to limit further host range evolution. This work emphasizes the advantages of the use of different levels of specialization and the inclusion of occasional hosts to establish a more detailed scenario for the evolution of this kind of ecological association.