The European Pharmacogenetic Opioid Study (EPOS) of a large series of European cancer patients treated with opioids was carried out to assess the influence of genetics on cancer pain relief. As response to opioid therapy was associated with the patients' country of origin, we tested whether population stratification might represent a confounding factor in the analysis of genetic control of response to opioid therapy. From the whole EPOS series representing 2294 patients' genotypes for 379 single-nucleotide polymorphisms (SNPs), we extracted 117 autosomal SNPs with minor allele frequency>0.28 to obtain highly informative genetic markers, and analyzed the SNPs in 1724 individuals showing <20% missing genotypes. Use of the AWclust program to detect clusters of genetically related individuals in the EPOS series showed that the 117-SNP panel distinguished four main European subgroups statistically associated with ethnicity, but not with country of origin or with the pain relief phenotype. Subethnic European groups of genetically related individuals exist that can be correctly identified using an ∼100-SNP panel. Such genetic clustering may control for admixture in association studies and may allow discrimination between genetic and environmental effects on phenotypes showing association with country of origin, as in the case of pain relief.