Subjects in five stress groups were threatened with electric shock while subjects in a sixth group were not. In one of the stress groups subjects were encouraged to misattribute (i.e., project) their feelings from the threat of shock to the experimenter instead of to the shock, while subjects in the remaining stress groups were not encouraged to misattribute threat-produced affects to the experimenter. Physiological and self-report measures indicated that subjects who misattributed their anxiety and nervousness to the experimenter were just as distressed by the veridical source of arousal as were subjects who did not misattribute. The implications of these results for attribution theory and the hypothesized stress-reducing effectiveness of complementary projection were discussed. (Author)