1. Interaction between alcohol and bretazenil (a benzodiazepine partial agonist in animals) was studied with diazepam as a comparator in a randomized, double-blind, placebo controlled six-way cross over experiment in 12 healthy volunteers, aged 19-26 years. 2. Bretazenil (0.5 mg), diazepam (10 mg) and matching placebos were given as single oral doses after intravenous infusion of alcohol to a steady target-blood concentration of 0.5 g l-1 or a control infusion of 5% w/v glucose at 1 week intervals. 3. CNS effects were evaluated between 0 and 3.5 h after drug administration by smooth pursuit and saccadic eye movements, adaptive tracking, body sway, digit symbol substitution test and visual analogue scales. 4. Compared with placebo all treatments caused significant decrements in performance. Overall, the following sequence was found for the magnitude of treatment effects: bretazenil+alcohol > diazepam+alcohol > or = bretazenil > diazepam > alcohol > placebo. 5. There were no consistent indications for synergistic, supra-additive pharmacodynamic interactions between alcohol and bretazenil or diazepam. 6. Bretazenil with or without alcohol, and diazepam+alcohol had marked effects. Because subjects were often too sedated to perform the adaptive tracking test and the eye movement tests adequately, ceiling effects may have affected the outcome of these tests. 7. No significant pharmacokinetic interactions were found. 8. Contrary to the results in animals, there were no indications for a dissociation of the sedative and anxiolytic effects of bretazenil in man.