In numerous population genetic and disease association studies decisions about the ancestry of polymorphic alleles are often made based on the relative frequency of the alleles in the extant populations with the most frequent allele being deemed as ancestral. However, the frequency of an allele in a population is generally not a perfect indicator of its ancestral status. A more accurate method to assess ancestral/derived status of polymorphic alleles involves identification of shared alleles between species. We used this strategy to examine genomic regions homologous to several human polymorphisms in four species of non-human primates. Cross species polymerase chain reaction (CS-PCR), with primers designed from human sequence, was used to investigate regions of interest. Nineteen polymorphisms at six loci (DRD2, HOXB@, PAH, D4S10, RBP3, and RET) were examined either by restriction fragment length analysis of PCR products (PCR-RFLP) or by direct sequencing. At seventeen of the eighteen PCR-RFLPs, non-human primates were monomorphic and identical to each other for either lack of restriction enzyme site or presence of the site. Thus, at these seventeen polymorphic sites the shared alleles are most likely to be the ancestral ones in humans. In several cases we have used sequence data to further demonstrate that the nucleotide at the site of the polymorphism is conserved between species confirming the hypothesis of a single ancestral allele. However, not all human alleles can be simply resolved into ancestral and derived; sequence data from one PCR-RFLP (in an intron of the PAH locus) and a single strand conformational polymorphism (SSCP) in the 3' untranslated region (UTR) of the DRD2 gene illustrate this point.