Forty-two undergraduates were classified into High-High Amptitude Group (H-H), High-Low Amptitude Group (H-L), and Low-Low Amptitude Group (L-L), according to the change patterns of their GSR waves. They were asked to reiterate classical Chinese poems and to take "subtests" of Raven's Progressive Matrices. Their lowest absolute skin resistance and scores on Raven's "subtests" were recorded before and after their performance being observed by their classmates. The subjects' perception of stressors was assessed by a rating scale of semantic differential type, with sixteen school-life events as its concepts, and with "Magnitude", "Controllability", and "Stability" as its concepts. Data thus obtained were analyzed by MANOVA, MANOCOVA, discriminant analysis and classification statistics. The results were as follows: (1) So far as their responses to psychosocial stressors were concerned, there was no significant difference among H-H, H-L, and L-L groups, though the stress was proved to be strong enough. (2) There were only 50% correct classifications. (3) There were no significant discriminant functions with which the H-H, H-L, and L-L groups may best be classified. Thus, it was concluded that physiological indices, such as change pattern of GSR waves, may not be used as a good measure of one's personality.