Genetic polymorphisms of human arylamine N-acetyltransferase 2 (NAT2) are responsible for interindividual variation in the acetylation of numerous drugs and in the transformation of aromatic and heterocyclic amines into carcinogenic intermediates. Although large interethnic variability in the frequency for NAT2 variant alleles has been reported, comparison of allele frequencies is hampered by differences in the criteria for the assignment of allelic variants. To avoid such sources of bias, in this review we analyze the occurrence of both interethnic and intraethnic variability for the seven commonest single nucleotide polymorphisms (SNP) in the NAT2 gene by using raw SNP data instead of inferred haplotypes. Besides the large interethnic variability observed for all SNPs except C282T, intraethnic variability for NAT2 SNPs was identified for the SNPs G191A among Caucasians (p < 0.0001), T341C among Oriental (p < 0.001) or African individuals (p < 0.012), C481T among Oriental (p < 0.001) or African individuals (p < 0.001), and G590A among Oriental individuals (p < 0.001). In contrast, no major intraethnic differences were identified for the SNPs C282T, A806G or G857A. Intraethnic variability may have relevant clinical implications. For instance, case-control NAT2 studies should not be extrapolated from one Oriental population to another. Nonsynonymous SNPs occur in 32% of alleles in Japanese individuals and in 47% of alleles in Chinese individuals, therefore the frequency of adverse effects and cancer related to slow acetylation is expected to be higher in individuals with Chinese descent than in those with Japanese descent. Intraethnic variability reinforces the need for proper selection of control subjects and points against the use of surrogate control groups for studies involving association of NAT2 alleles with adverse drug effects or spontaneous diseases.