Comparisons of the suitability of different tree species as food for the gypsy moth,Lymantria dispar(L.), have usually been based on measurements of larval growth on foliage samples taken from few tree individuals at single field sites. By rearing gypsy moth larvae on five tree species (three preferred [all oaks] and two marginal) at sites located in three physiographic regions, we tested whether differences in suitability among tree species can vary from location to location. Relative suitabilities did not vary among locations for weight of female pupae or development time of male larvae, but there were some differences for weight of male pupae and development time of female larvae. These tree species × location interactions, however, accounted for little of the overall variability in gypsy moth pupal weight and development time on the three oak species. The largest sources of variation, aside from sex, were differences among larvae on the same tree and differences among trees of the same species. Some factors that were controlled for in this study, such as tree size (trunk diameter was positively correlated with weights of male and female pupae), phenology, extent of defoliation, and genetic differences among gypsy moth populations, may contribute to differences in the relative suitability of tree species among naturally occurring sites of infestation. Nevertheless, results of this experiment suggest that in studies of relative suitability, sampling replicate trees within species is more crucial than including multiple study sites.