Ancient DNA provides a potentially revolutionary way to study biological relationships in prehistoric populations, but genetic patterns are complex and require careful interpretation based on robust, well-tested models. In this study, nuclear and mitochondrial markers were compared in the Yanomamö, to assess how well each data set could differentiate among closely related groups. The villages selected for the study share a recent fission history and are closely related to each other, as would likely be the case among prehistoric peoples living in the same valley or region. The Yanomamö generally practice village-level endogamy, but some migration and gene flow are known to occur between villages. Nuclear and mitochondrial DNA data were compared using F-statistics and genetic distance analyses. The nuclear data performed as expected, males and females from the same village were similar, and the villages were genetically distinct, with the magnitude of genetic differences correlated with historical relationship. However, mtDNA analyses did not yield the expected results. The genetic distances between villages did not correlate with historical relationship, and the sexes were significantly different from each other in two villages. Both the Lane and Sublett and the Spence methods, used to test for archaeological residence patterns, were consistent with endogamy. Hence, ancient DNA can, in principle, provide us with a unique opportunity to study genetic structure and gene flow in archaeological populations. However, interpretations, particularly those based on single loci such as mitochondrial DNA, should be cautious because sex-specific migration and sampling issues may have dramatic effects. Am J Phys Anthropol 117:246–259, 2002. © 2002 Wiley-Liss, Inc.