The kinetic and dynamic interaction of caffeine and zolpidem was evaluated in a double-blind, single-dose, six-way crossover study of 7.5 mg zolpidem (Z) or placebo (P) combined with low-dose caffeine (250 mg), high-dose caffeine (500 mg), or placebo. Caffeine coadministration modestly increased maximum plasma concentration (C(max)) and area under the plasma concentration-time curve of zolpidem by 30-40%, whereas zolpidem did not significantly affect the pharmacokinetics of caffeine or its metabolites. Compared to P+P, Z+P significantly increased sedation, impaired digit-symbol substitution test performance, slowed tapping speed and reaction time, increased EEG relative beta amplitude, and impaired delayed recall. Caffeine partially, but not completely, reversed most pharmacodynamic effects of zolpidem. Thus, caffeine only incompletely reverses zolpidem's sedative and performance-impairing effects, and cannot be considered as an antidote to benzodiazepine agonists.