Following 49 h of sleep deprivation, 37 healthy males ingested either 2.1, 4.3, or 8.6 mg*kg-1 body weight of caffeine in a randomized double-blind design. Subjects were sleep deprived for additional 12 h and venous blood samples were collected intermittently. Caffeine and primary metabolite concentrations were determined by HPLC. Pharmacokinetics were computed using the Lagran computer program. The ratio of the primary human metabolite, paraxanthine, to caffeine was used to estimate the rate of metabolism. There was a significant (p < 0.05) and disproportional increase in the dose normalized caffeine AUC and a significant decrease in its clearance associated with increasing dose. In addition, the paraxanthine to caffeine ratio significantly decreased with an increase in dose, indicating that the rate of caffeine metabolism decreased. These results demonstrate that under the conditions of severe sleep deprivation caffeine exhibits dose-dependent pharmacokinetics. In addition, these results are consistent with a model of capacity-limited metabolism.