Eight patients in a cardiac rehabilitation program, after exposure to two psychological stressors approximately equivalent with respect to cardiovascular reactivity, were given nonconcurrent progressive muscle relaxation training and retested for reactivity. They were then provided with relaxation training concurrently with one of the stressors and exposed again to the two stressors. No significant effects for nonconcurrent progressive muscle relaxation training were detected. Concurrent training, in contrast, produced reductions in both systolic and diastolic blood pressure. Reductions resulting from training on the target stressor showed little tendency to generalize to the nontarget stressor; the discrimination was particularly well defined for systolic blood pressure. We conclude that muscle relaxation techniques are maximally effective in reducing reactivity to psychological stressors when relaxation training is provided concurrently with the stressor. Our findings further suggest that to inculcate the relaxation response reliably across different situations, specific training to enhance generalization may be needed.