A comparative study of N-acetyltransferase 2 (NAT2) genotyping and phenotyping (caffeine test method) was performed on 211 persons to elucidate apparent discrepancies in the assignment of NAT2*12 and NAT2*13 alleles which occur in the literature. The study used the standard procedures of genotyping (two PCR runs and application of seven restriction enzymes) and phenotyping (determination of the two caffeine metabolites 5-acetylamino-6-formylamino-3-methyluracil (AFMU) and 1-methylxanthine (1X)), as documented in detail and validated by the Deutsche Forschungsgemeinschaft. The data were consistent with an AFMU/1X molar ratio of 0.85 as cut-off point (antimode) between phenotypically slow and rapid acetylators. Under this provision, several R/S allele combinations did not comply, either fully or partly, with their associated phenotypes. In particular, there was a wide phenotypic overlap of the alleged rapid allele combination groups (i) NAT2*12A/*5A; NAT2*12C/*5D; NAT2*4/*5B, (ii) NAT2*13/*6B; NAT2*4/*6A, and (iii) NAT2*13/*7A; NAT2*4/*7B. These groups obviously contained both phenotypically rapid and slow acetylators. If one assumes that the presence of one "wild type" allele NAT2*4 defines a rapid acetylator the assignment of the alleles NAT2*12A, NAT2*12C, and NAT*13 as determinants of a rapid acetylator phenotype must be questioned. This refers in particular to the nucleotide changes A803G (NAT2*12A, NAT2*12C) and C282T (NAT2*13). Based on discussions in the literature and the data presented here, there is accumulating evidence that current assignments of the NAT2*12 and NAT2*13 alleles as determinants of a rapid acetylator state should be reconsidered.