The present study contrasted caffeine's effects on individuals who expect caffeine to stimulate them and those who do not. Secondly, whether a message that caffeine rather than placebo was administered would also affect these two groups of subjects differently was investigated. The study was conducted single-blind in a 2 × 2 × 2 mixed design. The between subjects factor was whether they expected caffeine to stimulate them (E+) or not (E−) according to their self reports obtained before the experiment began. The within subjects factors were message (told caffeine vs told placebo) and beverage type (given caffeine vs placebo). Sixteen subjects in each group (n=32) performed on signal detection, memory scanning and delayed free recall tasks following ingestion of either caffeinated or decaffeinated coffee on two sessions each, a total of four experimental sessions. On each session, subjects were given a message regarding their drink (told caffeine vs told placebo). However, on two sessions there was a mismatch between the message and drink given. For signal detection, performance under caffeine was better than placebo in the E+ but not the E− group. However, subjects in the E+ group did not benefit more than the E− group in either message condition. On memory scanning, detections and false alarms did not differ for either beverage, nor was there a differential finding in the E+ and E− groups. However, reaction time under caffeine condition was shorter. No effects of message were found. Caffeine and message also did not have any effect on performance on the delayed free recall task. The hypothesis that caffeine and message would affect E+ and E− subjects differentially was partly supported. Copyright © 2005 John Wiley & Sons, Ltd.