The assumption that EMG biofeedback cultivates an antistress response was tested under stress conditions while investigating the comparative efficacy of low versus high arousal treatment strategies. Biofeedback-assisted, cue-controlled relaxation training was used as the low arousal treatment strategy for half of the 20 normal subjects used in the study. The other half received a high arousal treatment strategy which used the same training in combination with an avoidance conditioning procedure. In this procedure mild electric shock was used as contingent aversive stimulation designed to reinforce relaxation responses. Both groups received four in-lab training sessions with a 4-day interim of home practice of cuecontrolled relaxation prior to the last in-lab training session. Pretraining assessment consisted of four 10-minute periods of alternating no-stress and stress conditions. Mild electric shock and loud tones were used as stressors. Posttraining assessment was identical to pre training except subjects employed self-directed, cue-controlled relaxation rather than self-directed relaxation based on instructions without training. Frontal EMG, subjective mental and muscle tension ratings, and behavioral observations of relaxation behavior served as dependent measures during pre- and posttraining assessment. EMG readings were used during in-lab training and the two subjective rating scales were used during home practice.