The study was undertaken to show that polymorphic isoniazid elimination in humans is trimodal; that the acetylator genotype and eliminator phenotype of the individual patient are concordant; and that the differences in the pharmacokinetic parameters of fast, intermediate, and slow eliminator subgroups are statistically significant. Sixty adult patients of both sexes and of mixed race with tuberculosis participated in the trial. The apparent elimination rate constant (k, h(-1)) and the area under the isoniazid concentration-time curve (AUC, mg/L/h), over the interval 2 to 6 h after oral isoniazid were determined in all patients; NAT2 allele composition was determined in 47 patients. Serum INH concentrations were determined by HPLC and genotypes by PCR/restriction enzyme analysis. Three eliminator phenotypes could be distinguished, and concordance between the phenotype and the genotype of the individual could be demonstrated. The isoniazid concentration-time profiles of the three eliminator subgroups were significantly different (p < 0.05). The NAT2*12A allele, which codes for fast acetylation, has a high frequency in the population studied, the intermediate acetylator genotype is constituted of codominant fast and slow alleles, and the distribution of phenotypes/genotypes in the population is consistent with Hardy-Weinberg predictions. The therapeutic implications of polymorphic isoniazid metabolism are discussed.