In a comparison among relaxation procedures, 32 college students were assigned to four group of equal size. Electromyographic response training was given with biofeedback (EMG training) on the forehead (frontal area) alone, on the frontal area, neck (sternomastoid), and forearm areas in succession, or on these three sites in conjunction with recorded relaxation instructions used at home. Relative to a control group, which received no training, the three biofeedback-trained groups maintained lower EMG levels on the frontal and sternomastoid sites, and the group provided with the relaxation instructions plus EMG training showed lower skin conductance levels. These patterns were generally maintained during the presentation of a stimulus (stressor) that ostensibly signaled an impending electric shock. Other measures, including peripheral temperatures and self-reported anxiety, also showed effects consistent with the stressor presentations but did not differentiate the groups. The results are discussed in terms of common clinical relaxation procedures, an alternative procedure for training several sites simultaneously, and implications for models of EMG training and arousal.