The ability of a drug to cause clinically significant drug-drug interactions due to direct or metabolism-dependent inhibition of cytochrome P450 (CYP) can generally be predicted from in vitro studies with human liver microsomes (HLM) or recombinant CYP enzymes, as recommended by the FDA and other regulatory agencies. This review highlights some examples of system-dependent inhibition of CYP and uridine diphosphate glucuronosyltransferase (UGT) enzymes. In the case of CYP enzymes, examples are presented where in vitro studies with HLM under-predict or over-predict the degree of inhibition observed in the clinic and where the correct prediction comes from studies with human hepatocytes. Studies with HLM under-predict the ability of gemfibrozil and bupropion to cause clinically significant inhibition of CYP2C8 and CYP2D6, respectively, and over-predict the ability of ezetimibe to cause clinically significant inhibition of CYP3A4. Gemfibrozil and bupropion represent examples of glucuronidation-dependent and reduction-dependent activation to metabolites that inhibit CYP2C8 and CYP2D6, respectively, whereas ezetimibe represents an example of glucuronidation-dependent protection against metabolism-dependent inhibition of CYP3A4. This article illustrates why, when drug candidates are extensively metabolized by non-CYP enzymes, it would be prudent to use human hepatocytes in addition to HLM or recombinant enzymes to evaluate their ability to inhibit CYP enzymes.