CYP3A activity in adults varies between individuals and it has been suggested that this has a genetic basis, possibly related to variant alleles in CYP3A4 and CYP3A5 genes. Accordingly, genotype-phenotype associations were investigated under constitutive and induced conditions. Midazolam's systemic and oral clearances, and the erythromycin breath test (ERBT) were determined in 57 healthy subjects: 23 (11 men, 12 women) European- and 34 (14 men, 20 women) African-Americans. Studies were undertaken in the basal state and after 14-15 days pretreatment with rifampin. DNA was characterized for the common polymorphisms CYP3A4*1B, CYP3A5*3, CYP3A5*6 and CYP3A5*7 by direct sequencing, and for exon 21 and exon 26 variants of MDR1 by allele-specific, real-time polymerase chain reaction. In 95% of subjects, the basal systemic clearance of midazolam was unimodally distributed and variability was less than four-fold whereas, in 98% of the study population, oral clearance varied five-fold. No population or sex-related differences were apparent. Similar findings were observed with the ERBT. Rifampin pretreatment markedly increased the systemic (two-fold) and oral clearance (16-fold) of midazolam, and the ERBT (two-fold) but the variabilities were unchanged. No associations were noted between these phenotypic measures and any of the studied genotypes, except for oral clearance and its fold-increase after rifampin. These were related to the presence of CYP3A4*1B and the inversely linked CYP3A5*3 polymorphism, with the extent of induction being approximately 50% greater in CYP3A5*3 homozygotes compared to wild-type subjects. In most healthy subjects, variability in intestinal and hepatic CYP3A activity, using midazolam as an in-vivo probe, is modest and common polymorphisms in CYP3A4 and CYP3A5 do not appear to have important functional significance.