Polygraphic monitoring of several physiological variables was done throughout an experiment investigating the effects of caffeine on mental performance. The experiment started with a mental maze learning task. Then the subjects were given the test beverages according to the group design (CC group (N = 16): 300 mg caffeine in decaffeinated coffee, DC group (N = 16): decaffeinated coffee, WW group (N = 8): warm water, and NB group (N = 8): no beverage). The experiment continued with a letter cancellation task which was followed by a second mental maze learning task. The caffeine treated subjects differed from the other groups by increased regularity of letter cancellation performance, as indicated by decreases in intraindividual variance. They also differed from the other groups by a slight but significant acrodermal vasoconstriction. No intergroup differences were obtained for mental maze learning, heart rate, respiration, muscle tension, and skin conductance. The results suggest therefore that the drug at this dose level improves behavioral routine and speed rather than cognitive functions and that the vegetative side effects are minimal.