Sixty male and 60 female introductory psychology student volunteers participated in the study. A Lafayette model 760-566 polygraph was used to record skin resistance responses and respiration. A total of 21 male and 23 female subjects were asked to write out in some detail a true, very embarrassing incident about which they would rather not tell anyone. A second set of subjects followed instructions that led them to be guilty or innocent of a mock murder. All subjects were reminded that during the polygraph examination they were to deny their involvement in the mock crime or the embarrassing incident. In this way, half of the subjects were deceptive and half were truthful about the events. The major analyses involved 2 x 2 x 2 MANOVA and univariate analyses on detection scores derived from the polygraph recordings of cardiovascular, respiration, and skin resistance measures. Gender, situation (mock crime, embarrassing events), and condition (innocent or guilty) were the independent variables. Findings show that deceptive subjects were relatively more reactive to event-relevant questions than control questions; whereas, innocent subjects were more reactive to control questions. With skin resistance response and cardiovascular measures, subjects examined by the police scored more toward innocence; whereas, those tested by laboratory investigators scored more toward guilty. Such a result could mean that laboratory investigators, when mistaken, would have a tendency to classify innocent people as guilty; and the police, when wrong, would tend to classify the guilty as innocent. 2 tables, 1 figure, and 23 references