BACKGROUND: Population stratification is the main source of spurious results and poor reproducibility in genetic association findings. Population heterogeneity can be controlled for by grouping individuals in ethnic clusters; however, in admixed populations, there is evidence that such proxies do not provide efficient stratification control. The aim of this study was to evaluate the relation of self-reported with genetic ancestry and the statistical risk of grouping an admixed sample based on self-reported ancestry. METHODS: A questionnaire that included an item on self-reported ancestry was completed by 189 female volunteers from an admixed Brazilian population. Individual genetic ancestry was then determined by genotyping ancestry informative markers. RESULTS: Self-reported ancestry was classified as white, intermediate, and black. The mean difference among self-reported groups was significant for European and African, but not Amerindian, genetic ancestry. Pairwise fixation index analysis revealed a significant difference among groups. However, the increase in the chance of type 1 error was estimated to be 14%. CONCLUSIONS: Self-reporting of ancestry was not an appropriate methodology to cluster groups in a Brazilian population, due to high variance at the individual level. Ancestry informative markers are more useful for quantitative measurement of biological ancestry.