Classical biological control could have a major environmental cost if introduced natural enemies colonize and disrupt native systems. Although quantifying these impacts is difficult for systems already colonized by natural enemies, the a priori condition for such impacts can be evaluated based on the extent to which exotics have acquired native hosts. We use native host records for exotic parasitoids introduced into North America for biological control to document the number of exotic species that have been recorded from at least one native insect species. We also evaluate the ability of six biological and ecological variables to predict whether or not a parasitoid will move onto natives. Sixteen percent of 313 parasitoid species introduced against holometabolous pests are known from natives. Further, the likelihood that a parasitoid had colonized native hosts was largely unpredictable with respect to the independent variables. We conclude that given the quality of the data available either now or in the foreseeable future, coupled with inherent stochasticity in host shifts by parasitoids, there are no rules of thumb to assist biological control workers in evaluating if an introduced parasitoid will colonize native insect communities.