The North American species of Celtis have been believed on anecdotal grounds to hybridize commonly in nature, and an understanding of the nature and frequency of natural hybridization among species of Celtis is necessary for understanding phylogenetic relationships within the genus. Extensive natural introgression has been strongly suspected between two species with ranges that overlap broadly in Texas, Celtis laevigata and C. reticulata. Statistical study of morphological and AFLP data was undertaken to test the extent and genetic consequences of natural hybridization between them. Results indicate that there is little or no natural hybridization between these species, contradicting past conclusions. The variability and apparent intergradation between the species that led to the hypothesis of introgression are apparently due to high levels of shared ancestral polymorphism at many loci, and exceptionally high levels of position-associated variation in leaf morphology within individual trees.