Human molars exhibit varying shapes when viewed from the occlusal surface. Available methods for quantifying molar occlusal shape have historically been confined to qualitative descriptions. The present study utilized geometric morphometric analyses to capture molar shape as defined through relative cusp locations. Cusp apices of maxillary and mandibular first and second molars were digitized from 190 American Blacks and Whites to estimate biological affinity through the shape of relative cusp locations. The coordinate data were subjected to a Generalized Procrustes Analysis to generate Procrustes coordinates and calculate centroid sizes. Procrustes coordinates were then subjected to a principal component analysis to examine the direction and magnitude of shape change inherent in the sample. Centroid size and major shape component group means were compared with t-tests. Interlandmark distances were then calculated from the raw coordinate information and also subjected to a principal components analysis. Procrustes coordinates and the principal components derived from them with and without centroid size, along with the interlandmark distances and the principal components derived from them, were each subjected to a discriminant function analysis to examine which methods yielded the highest correct classification between population groups. Total correct classifications ranged from 62.7% to 87.9% depending on the variables forward stepwise selected for each analysis. Using a combination of the second maxillary molar and first mandibular molar yielded the most optimistic results and corroborates theoretical models of molar development. Am J Phys Anthropol 153:269–279, 2014. © 2013 Wiley Periodicals, Inc.