Biological control is sometimes the only feasible option for controlling widespread invasive weeds, and the introduction of safe and effective agents are the primary goals. Insect herbivores commonly exhibit local adaptations to their host(s), and, thus, knowledge of what constitutes a population and how differentiated populations are can be useful in 1) determining how large an area should be sampled for host specificity testing so that the variation within the population gets tested; 2) delimiting the area suitable for collection of source material once an organism is deemed safe and effective; and 3) locating potentially distinct populations for additional testing, which might be useful if, for example, the tested population is sufficiently host specific but not highly effective. Here we present data on the population structure of Ceutorhynchus scrobicollis, a weevil proposed for the control of garlic mustard, Alliaria petiolata. We used Amplified Fragment Length Polymorphisms (AFLPs) to evaluate relationships among populations of C. scrobicollis collected from 11 sites in Europe. We found significant differentiation among sites that accounted for approximately 30% of the genetic variation. Over all sites there was a strong pattern of isolation by distance, however, within the vicinity of Berlin this pattern did not hold. Host specificity tests have been performed on individuals from the Berlin region, and our data suggest that these individuals comprise a single population that harbors substantial neutral genetic variation. More distant sites were significantly differentiated from the Berlin region, however, and should be tested prior to use as a weevil source.