The subjects were 24 student actors, aged 19 to 52 years and trained in the Stanislavsky method of acting. The 'guilty' group contained six males and six females and the 'innocent' group contained seven males and five females. All subjects were motivated to appear innocent so as to obtain the monetary bonus and exhibit their acting ability. Two versions of the control question detection of deception test were employed: one in which subjects verbally answered immediately following each question delayed their verbal answer for 8 seconds following each question (DAT). The delayed answer technique allowed the separate measurement of physiological responses to the questions and the answers. Skin resistance responses, cardiovascular changes, and respiration were recorded. The findings showed that while the decisions were 100 percent correct with guilty subjects, both the classification data and the quantitative data indicated less success with innocent subjects. A total of 17 percent of the subjects classified as 'deceptive' were truly innocent. Thus, the PDD control question technique may be more useful in 'clearing' truly innocent subjects than in conclusively identifying truly guilty subjects. The DAT results indicated that the physiological changes which occurred in response to the questions were more valid indicators of deception and nondeception than those which occurred in response to the verbal answer. However, the responses to the questions did not usually fully recover to initial baseline levels before the subjects answered. Nevertheless the DAT results were consistent with the IAT results that reactions were observed shortly after the question started and prior to the subject's answer. Thus, attempted deception could be detected even if the subject remained silent. As for the individual response system, the skin resistance responses proved to be the most useful and consistent on an individual basis. The cardiovascular measure was useful, but not as consistent on an individual basis, and the respiratory measures (cycle time and amplitude) were unreliable. Statistical data and about 20 references are included. (Author abstract modified)