As a direct-acting inhibitor of CYP2C19 in vitro, lansoprazole is more potent than omeprazole and other proton pump inhibitors (PPIs), but lansoprazole does not cause clinically significant inhibition of CYP2C19 whereas omeprazole does. To investigate this apparent paradox, we evaluated omeprazole, esomeprazole, R-omeprazole, lansoprazole, and pantoprazole for their ability to function as direct-acting and metabolism-dependent inhibitors (MDIs) of CYP2C19 in pooled human liver microsomes (HLM) as well as in cryopreserved hepatocytes and recombinant CYP2C19. In HLM, all PPIs were found to be direct-acting inhibitors of CYP2C19 with IC(50) values varying from 1.2 μM [lansoprazole; maximum plasma concentration (C(max)) = 2.2 μM] to 93 μM (pantoprazole; C(max) = 6.5 μM). In addition, we identified omeprazole, esomeprazole, R-omeprazole, and omeprazole sulfone as MDIs of CYP2C19 (they caused IC(50) shifts after a 30-min preincubation with NADPH-fortified HLM of 4.2-, 10-, 2.5-, and 3.2-fold, respectively), whereas lansoprazole and pantoprazole were not MDIs (IC(50) shifts < 1.5-fold). The metabolism-dependent inhibition of CYP2C19 by omeprazole and esomeprazole was not reversed by ultracentrifugation, suggesting that the inhibition was irreversible (or quasi-irreversible), whereas ultracentrifugation largely reversed such effects of R-omeprazole. Under various conditions, omeprazole inactivated CYP2C19 with K(I) (inhibitor concentration that supports half the maximal rate of inactivation) values of 1.7 to 9.1 μM and k(inact) (maximal rate of enzyme inactivation) values of 0.041 to 0.046 min(-1). This study identified omeprazole, and esomeprazole, but not R-omeprazole, lansoprazole, or pantoprazole, as irreversible (or quasi-irreversible) MDIs of CYP2C19. These results have important implications for the mechanism of the clinical interaction reported between omeprazole and clopidogrel, as well as other CYP2C19 substrates.