Host range is a fundamental aspect of host-pathogen population dynamics and is a vital determinant of the risks associated with the use of a pathogen for biological control. Yet even for widely studied baculoviruses we often do not know which hosts they infect under natural conditions. Baculoviruses are primarily pathogens of Lepidoptera and Hymenoptera and exhibit great variation in host range. Some are apparently monospecific, whereas others are able to infect numerous hosts from diverse families. In this report, I review host range data, concentrating on two generalist viruses, Autographa californica (Speyer) multicapsid nucleopolyhedrovirus (AcMNPV) and Mamestra brassicae L. multicapsid NPV (MbMNPV). To my knowledge, 145 and 78 host species have been assayed with AcMNPV and MbMNPV, respectively. I examined susceptibility of these species according to various life history parameters using a phylogenetic analysis. Host susceptibility was poorly predicted by taxonomic relationships. Host susceptibility was correlated with some life history traits; of the species tested, a significantly higher proportion of polyphagous species and pest species were susceptible to AcMNPV compared with mono- or oligophagous species or nonpest species, respectively. A higher proportion of multivoltine species were susceptible to MbMNPV compared with univoltine species, although the opposite trend was shown for AcMNPV. Providing convincing explanations for these patterns is problematic and highlights the inadequacies in our understanding of the ecology of baculoviruses. Some possible explanations are discussed. I suggest that current patterns of host susceptibility may, in part, reflect an ongoing evolutionary arms race between numerous host species and each baculovirus, in which the virus repeatedly colonizes new host populations and new species while others evolve resistance.