Phenytoin, an anticonvulsant, exhibits nonlinear pharmacokinetics with large interindividual differences. Because of its small therapeutic range with the risk of therapeutic failure or adverse drug effects in susceptible persons, therapeutic drug monitoring is frequently applied. The interindividual differences in dose response can partially be explained by known genetic polymorphisms in the metabolic enzyme CYP2C9 but a large deal of individual variability remains still unexplained. Part of this variability might be accounted for by variable uptake of phenytoin, which is a substrate of p-glycoprotein, encoded by the human MDR1 gene. We evaluated, whether phenytoin plasma levels correlate with a polymorphism in the MDR1 gene, C3435T, which is associated with intestinal PGP activity. Genotyping and analyses of plasma levels of phenytoin and metabolites in 96 healthy Turkish volunteers showed that the MDR1C > T3435 polymorphism affects phenytoin plasma levels (P = 0.064) and the metabolic ratio of p-HPPH vs phenytoin (MDR1*TT genotype, P = 0.026). The MDR1*CC genotype is more common in volunteers with low phenytoin levels (P < or = 0.001, chi2 test). A combined analysis of variable alleles of CYP2C9, 2C19 and MDR1 revealed that the number of mutant CYP2C9 alleles is a major determinant, the number of MDR1*T alleles further contributes to the prediction of phenytoin plasma levels and CYP2C19*2 does not explain individual variability. The regression equation that fitted the data best included the number of mutant CYP2C9 and MDR*T alleles as predictory variables and explained 15.4% of the variability of phenytoin data (r2 = 0.154, P = 0.0002). Furthermore, analysis of CYP2C9 and MDR1 genotypes in 35 phenytoin-treated patients recruited from therapeutic drug monitoring showed that combined CYP2C9 and MDR1 analysis has some predictive value not only in the controlled settings of a clinical trial, but also in the daily clinical practice.