Two experiments (N = 56) investigated the relationship between subjects' expectancies concerning the effect of caffeine on a motor skill, and the type of placebo response. Male subjects were assigned to four groups. Three groups expected to receive caffeine but received a placebo. Prior to the placebo, two of the groups received information about the effect of caffeine on a motor skill task which led one group E(+) to expect enhanced performance, and the other E(-) to expect impairment. The third placebo group received no information E(?). A control group E(0) received no beverage, so neither caffeine nor any effect on performance was expected. The expected type of effect predicted the type of placebo response displayed. Group E(+) displayed greater improvement under placebo than did group E(0), and group E(-) performed more poorly than those in group E(0). No placebo response was observed in group E(?). Placebo effects on mood were correlated with subjects' predictions about the effect of caffeine on mood. The role of expectancies in response to placebos and psychoactive drugs is discussed.