Sixty-five college students served as the subjects to examine the effects of loud noise (70 to 80 dB) on the behaviors of problem solving and dichotic listening. They were first 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. Members of each group were then randomly assigned to Noise Groups and Control Groups to accept different treatments. The subjects were instructed to listen carefully to the contents of two stories told simultaneously by a man and a woman (Two Story task), to listen carefully to the word-pairs presented simultaneously to left ear and to right ear (Dichotic Listening task), and to solve problems consisted of anagrams, cryptarithmetic materials, etc. (Problem Solving task). Bursts of noise of 70 to 80 dB(A) were presented to the Noise Group during the experiment. Data thus obtained were analyzed by MANOV A and ANOV A. The findings were as follows: (1) There were significant differences between Noise Groups and Control Groups on the scores of all three kinds of task. The loud noise apparently had harmful effect on the behaviors of problem solving and dichotic listening. (2) There was no significant difference among H-H Group, H-L Group, and L-L Group. Thus it seemed impossible to predict from GSR patterns the behaviors of problem solving and dichotic listening. (3) Among the sixteen personality factors in Cattell's 16 PF, five factors were found to have significant correlation with the classification of GSR patterns when triserial correlation method was applied.