The present study investigated the dose-effect relationship of caffeine on mental performance using a caffeine-sensitive rapid information processing task (RIP) in a pre/post cross-over design. Twenty female nonsmoking regular coffee drinkers received 0, 1.5, 3 and 6 mg/kg caffeine in a balanced order and the measurements were extended to cardiovascular, EEG and mood parameters. Surprisingly, the dose-effect curves for the different parameters were rather heterogeneous. Whereas increasing effects with increasing caffeine doses were observed for alpha- and beta-EEG frequencies, anxiety, wakefulness, and some coffee ratings, negative dose-effect relationships were obtained for RIP processing rate and blood pressure. No apparent dose-effect relationships were seen for reaction time and motor activity. Thus, it was concluded that the dose-response relationships are rather shallow and heterogeneous and that the different parameters have different ranges in which they are sensitive to caffeine. The caffeine doses which might have beneficial behavioral effects are at the lower end of the tested dose range and comparable to those found in caffeine-containing beverages.