A new method of scoring dental microscopic use wear, initially developed for and applied to extant and extinct ungulates, is here applied to primates, and the efficacy of the method as a tool for diagnosing diet in both ungulates and primates is established. The method employs standard refractive light microscopy instead of scanning electron microscopy (SEM), and all use-wear features are counted or scored under low magnification (35X). We use measurement systems analysis (variance components analysis of sources of measurement error) to evaluate the consistency and reproducibility of measurements using this method. The method is shown to have low intra- and inter-observer measurement error, and to effectively distinguish among graminivores, folivores, and frugivores. It can also be used to identify seed predators and to diagnose hard-object feeding. The method is also shown to be robust to the selection of measurement site; it works equally well when applied to upper or to lower molars. Finally, we use analysis of variance to examine the consistency of the signals across mammalian orders, and discriminant function analysis to develop dietary diagnostic tools for a set of "classified" primates with known diets. We test the success of these tools not merely by examining their a posteriori classification "success," but by using them to construct predicted dietary profiles for a sample of unclassified extant primate species, again with known diets. (C) 2004 Elsevier Ltd. All rights reserved.