Cultured hepatocytes from newborn human (three samples), adult human (eight samples) and adult rat livers were used to study the metabolism of theophylline and caffeine, two drugs of which the metabolic pathways are known to be cytochrome P-450-dependent. Known metabolic pathways of caffeine in vivo were qualitatively maintained. However, only the primary metabolites were formed through oxidative N-demethylation giving theophylline, paraxanthine and theobromine and, through C-8 hydroxylation, giving 1,3,7-trimethyluric acid and a ring-opened compound the 6-amino-5[N-formylmethylamino]1,3-dimethyl uracil. The ratio of the three dimethylxanthine metabolites was dependent upon the species (human, rat), development stage (newborn, adult) and environmental factors. Similarly, theophylline was metabolized as in vivo by the demethylation pathway giving, preferentially, 3-methylxanthine and not 1-methylxanthine, and by a C-8 oxidation giving 1,3-dimethyluric acid. In newborn hepatocytes, all pathways were absent except the well-known methylation to caffeine. Moreover, such a methylation also occurred in adult human hepatocytes. This result was explained by the very low metabolic capacity of cultured cells, allowing the detection of only direct metabolites. Indeed, the overall biotransformation of both the methylxanthines by primary cultures of hepatocytes was remarkably weak, confirming previous studies with liver microsomal incubations. Thus the metabolism rate did not exceed about 30 nmoles/10(6) cells/24 hr in human adults, except for two subjects which were characterized by an extensive metabolism and a different metabolic profile. These two subjects were probably induced in vivo by environmental compounds. Both quantitative and qualitative data obtained from this study were roughly correlated with other in vivo and in vitro studies. Overall the experimental model of cultured human hepatocytes was shown to be capable of assessing the metabolic profile of two methylxanthines which is in agreement with the situation encountered in vivo. This example suggests that a breakthrough may be brought in new drugs development by the predictability from human hepatocyte culture model to the in vivo human situation.