Historically, chemical ecologists assumed that cucurbitacin feeding and sequestration in rootworm leaf beetles are remnants of an ancient association between the Luperini (Coleoptera: Chrysomelidae; Galerucinae) and Cucurbitaceae (ancestral host hypothesis). Under this premise, rootworms that do not develop on cucurbits but undergo pharmacophagous forays for cucurbitacins are thought to do so to supplement novel host diets that lack these bitter compounds. The ancestral host hypothesis is supported by studies of pyrrolizidine alkaloid pharmacophagy in Lepidoptera, but has not been subjected to phylogenetic analysis within the Luperini. New evidence that this feeding behavior is better correlated with an adult affinity for pollen than with larval host offers the possibility that Old and New World rootworm species with an affinity for cucurbitacins converged on this behavior through apomorphic taste receptor modifications (loose receptor hypothesis). Here, the authors test the monophyly of cucurbitacin feeding and pharmacophagy within the Luperini by using nuclear and mitochondrial sequence data to infer phylogenetic relationships among 138 taxa representing tribes of the Galerucinae and subtribes of the Luperini. The resulting phylogenetic hypothesis is mostly concordant with existing tribal and subtribal delineations within the Subfamily Galerucinae sensu stricto (Galerucinae not including the flea beetles). The establishment of ancestry among the subtribes of the Luperini refutes the monophyly of cucurbitacin feeding and cucurbit specialization, with the New World Diabroticina being paraphyletic to the Old World Aulacophorina and cosmopolitan Luperina. This reconstructed paraphyly of the rootworms, from a modest taxon sampling of two disparate molecules subjected to a variety of statistical optimality criterion, suggests that revision may be needed at the tribal level within the Galerucinae sensu stricto. These data unambiguously support the convergent evolution of cucurbitacin feeding and pharmacophagy in rootworms and are inconsistent with the ancestral host hypothesis.