In vitro metabolic models using human liver microsomes can be applied to quantitative prediction of in vivo drug interactions caused by reversible inhibition of metabolism. One approach utilizes in vitro Ki, values together with in vivo values of inhibitor concentration to forecast in vivo decrements of clearance caused by coadministration of inhibitor. A critical limitation is the lack of a general scheme for assigning intrahepatic exposure of enzyme to inhibitor or substrate based only on plasma concentration; however, the assumption that plasma protein binding necessarily restricts hepatic uptake is not tenable. Other potential limitations include: flow-dependent hepatic clearance, "mechanism-based" chemical inhibition, concurrent induction, or a major contribution of gastrointestinal P450-3A isoforms to presystemic extraction. Nonetheless, the model to date has provided reasonably accurate forecasts of in vivo inhibition of clearance of several substrates (desipramine, terfenadine, triazolam, alprazolam, midazolam) by coadministration of selective serotonin reuptake-inhibitor antidepressants and azole antifungal agents. Such predictive models deserve further evaluation, since they may ultimately yield more cost-effective and expeditious screening for drug interactions, with reduced human drug exposure and risk.