Data on both the incidence of slow acetylator phenotype of probe drugs isoniazid, sulfadimidine or sulfamethazine, caffeine and dapsone in mainland or overseas Chinese, and the distribution of NAT2 genotypes and the frequency of NAT2 alleles in the Chinese populations were summarized and reanalysed using a meta-analysis method. Frequency of the slow acetylator phenotype in 3516 healthy Han Chinese gave an overall mean of approximately 19.9 +/- 4.0%, with the range of the combined data being between 15.8% and 25.5%. In addition, frequencies of the slow acetylator phenotype differ between the different minorities in Chinese populations and the range was between 3.2% and 50.6%, with a mean value of 20.6 +/- 12.9% in a total of 1842 individuals from 17 Chinese minorities. In addition, there was no significant heterogeneity in overseas Chinese between the probe drugs isoniazid and sulfadimidine or sulfamethazine (chi 2 = 5.97, df = 4; p > 0.05), and the mean value of slow acetylator phenotype incidence was 24.5% (119/485; 95% CI: 20.7-28.3%), consistent with that of the native Chinese. As expected, frequency of the slow acetylator genotypes in Chinese populations was 25.4% (112/441; 95% CI: 21.3-29.5%), which was in accordance with that of the slow acetylator phenotype in native or overseas Chinese. For all genotypes, *4/*4 (29.9%, 132/441), *4/*6A (27.4%, 121/441), *4/*7A (12%, 53/441) and *6A/*6A (11.3%, 50/441) occupied 80.6%, but *5A/*7A (0.2%, 1/441), *5A/*5A (1.1%, 5/441) and *7A/*7A (1.8%, 8/441) were not frequently found. From this report, the genotype frequencies of homozygous rapid acetylator, heterozygous rapid acetylator, and homozygous slow acetylator were found to be 0.299 (132/441), 0.447 (197/441) and 0.254 (112/441), respectively. Furthermore, both *4 (52.3%; 95% CI: 49-56%) and *6A (30.5%; 95% CI: 28-34%) were major NAT2 alleles, while *7A (11.2%; 95% CI: 9-13%) and *5A (6%; 95% CI: 4-8%) were uncommonly present. Frequency of the mutant alleles was observed at 0.477 (421/882 alleles). The *7A constituted 23.5% t(99/421) of slow acetylator alleles in Chinese populations, showing that this point mutation exists not only in Oriental or Asiatic, but also in Chinese populations. According to the Hardy-Weinberg equilibrium, in the phenotyped Chinese populations, the mean estimate of predicted allelic frequencies of the genotypes RR, Rr, and rr was 0.294, 0.496, and 0.210 for the Chinese, and the expected frequency of the deficient gene r was 0.458. By comparison, the predicted values are in complete agreement with the observed ones. In conclusion, this meta-analysis determined the accurate population frequencies of phenotype and genotype of the NAT2 genetic deficiency in healthy Chinese subjects.