The problem of the uncooperative experiment arises with the use of human subjects. Evidence shows that typical volunteer subjects have the following characteristics: better education, higher paying jobs, greater need for approval, lower authoritarianism, higher I.Q. score, and better adjustment to personal questions than nonvolunteers. Data also suggest that volunteers are more sociable, arousal seeking, younger, firstborn, and more unconventional than nonvolunteers. How representative, then, can volunteers be? Influences of experimenter on subject must also be considered. An experimenter can unknowingly communicate expectancies through transmission of cues. Professional experimenters should be carefully trained in the detection and control of artifact and expectancy-demand characteristics of experiments. There has also been discussion of using a more natural setting than the laboratory. Individual human characters and differences may lead to confusing results in an experiment despite all efforts to control behavior. (PB)