To determine whether caffeine antagonizes the decremental effects of triazolam and zopiclone on human performance, oral single doses of 0.250 mg triazolam, 7.5 mg zopiclone, or respective placebos, with and without 300 mg caffeine, were given to parallel groups of student volunteers in two double-blind studies. Objective tests and subjective visual analogue ratings were done at baseline and 30 min. and 90 min. after the intake. In Study I, triazolam produced drowsiness at 30 min. but did not differ from the placebo in other tests. Caffeine induced alerting effects in various tests and differed from triazolam in some (digit substitution, drowsiness, calmness, mental slowness) but not all variables measured. Caffeine and triazolam were interpreted as being antagonists. In Study II, zopiclone impaired digit substitution and flicker fusion, produced exophoria and lowered systolic blood pressure. Caffeine differed from zopiclone in several test functions, but it also differed from caffeine + zopiclone whereas zopiclone differed from caffeine + zopiclone only in two tests (Maddox wing, systolic blood pressure). Thus, zopiclone counteracted the effects of caffeine more easily than caffeine counteracted the decremental effects of zopiclone. We conclude that triazolam may not differ importantly from diazepam as regards their antagonism towards caffeine, whereas further research on the antagonism between zopiclone and caffeine needs to be done.