The cytochrome P450 enzyme CYP1A2 is crucial for the metabolism of many drugs, for example, tizanidine. As the effects of several non-steroidal anti-inflammatory drugs (NSAID) and female sex steroids on CYP1A2 activity in vitro are unknown, their effects on phenacetin O-deethylation were studied and compared with the effects of model inhibitors in human liver microsomes, followed by prediction of their interaction potential with tizanidine in vivo. In vitro, fluvoxamine, tolfenamic acid, mefenamic acid and rofecoxib potently inhibited CYP1A2 [the 50% inhibitory concentration (IC(50)) < 10 microM]. Ethinyloestradiol, celecoxib, desogestrel and zolmitriptan were moderate (IC(50) 20-200 microM), and etodolac, ciprofloxacin, etoricoxib and gestodene weak inhibitors of CYP1A2 (IC(50) > 200 microM). At 100 microM, the other tested NSAIDs and steroids inhibited CYP1A2 less than 35%. Pre-incubation increased the inhibitory effects of rofecoxib, progesterone and desogestrel. Using the free portal plasma inhibitor concentration and the competitive inhibition model, the effect of fluvoxamine and the lack of effects of tolfenamic acid and celecoxib on tizanidine pharmacokinetics in human beings were well predicted. However, the effects of ciprofloxacin, rofecoxib and oral contraceptives were greatly underestimated even when the predictions were based on their total portal plasma concentration. Besides rofecoxib, and possibly mefenamic acid, other NSAIDs were predicted not to significantly inhibit CYP1A2 in human beings. The type of enzyme inhibition, particularly metabolism-dependent inhibition, free inhibitor concentration and accumulation of the inhibitor into the hepatocytes should be considered in extrapolations of in vitro results to human beings.