The 10-year prognostic significance of psychosocial as well as medical risk factors was examined in 150 middle-aged Swedish men. Type A behavior was assessed by means of the Structured Interview; work demand, social support, and other psychosocial factors were registered through standardized questionnaires. The clinical investigation included a standard physical examination, a frontal and sagittal chest x-ray, fasting serum lipids, glucose, and urate, and a 24-hour ambulatory ECG monitoring. Thirty-seven men died during follow-up. Mortality was similar in men with Type A (24%) and Type B (22%) behavior. In multivariate analyses, lack of social support/social isolation was an independent mortality predictor in Type A, but not in Type B men. In both groups, a high frequency of ventricular ectopic beats on 24-hour ECG monitoring and a poor self-rated general health predicted mortality over the 10-year period. The 10-year mortality experience of socially isolated Type A men was 69% and that of socially integrated Type A men was 17% (p less than 0.05). The findings offer a possible explanation for the observed inconsistencies between intervention and follow-up studies of Type A behavior and coronary heart disease. It is suggested that an important effect of Type A modification programs is to increase the availability of social support. This could be the mechanism through which Type A modification exerts its main effects on cardiovascular health.