Despite the extensive use of ketoconazole as an index inhibitor of human cytochrome P450 3A (CYP3A) isoforms in vitro, literature reports of the quantitative inhibitory potency of ketoconazole are highly variable. In 51 published studies reporting 76 values of ketoconazole inhibition constants (K(i)) versus in vitro clearance of 31 different CYP3A substrates, the K(i) values ranged from 0.001 µM to 25 µM. The geometric mean was 0.1 µM (90% confidence interval: 0.07 to 0.15 µM), and the median was 0.08 µM. Even for one specific substrate metabolized to one specific metabolite (midazolam α-hydroxylation), variability was still extensive (K(i) range: 0.004-0.18 µM). Only about 20% of overall variability in K(i) was explained by a combination of incubation, duration, and microsomal protein concentration. The remaining variation is unexplained, but could be attributable to factors such as: in vitro clearance by non-CYP3A pathways; incorrect assignment of inhibition mechanism; and variable relative content of CYP3A4 and CYP3A5 in different microsomal preparations. However, the role of these factors still is not established. Until sources of variation are more clearly defined, variability can be minimized by use of low microsomal protein concentrations, short incubation periods, and data analysis procedures that use untransformed reaction velocities and inhibition models that allow for mixed competitive-noncompetitive mechanisms.