The relation of behavior (Type A or Type B) to the morbidity and mortality of coronary heart disease (CHD) is still debated. We studied the survival of 257 male patients with CHD from the initial, 8.5-year phase of the Western Collaborative Group Study to see whether behavior type--as assessed by a structured interview before the CHD event--was related to subsequent CHD mortality. Behavior type was not related to mortality in 26 patients who died within 24 hours of the coronary event. However, of the 231 patients who survived for 24 hours, the mortality rate associated with CHD among 160 Type A patients studied during an average 12.7 years was 19.1 per 1000 person-years. This was unexpectedly lower than the corresponding rate of 31.7 among 71 Type B patients who were followed for an average of 11.5 years (P = 0.04). In a proportional-hazards survival analysis, which controlled for variable follow-up time, the type of initial coronary event, and traditional risk variables, the relative CHD-associated mortality rate among Type A as compared with Type B patients was 0.58 (P = 0.03; 95 percent confidence interval, 0.35 to 0.96). The lower mortality among Type A subjects occurred in both younger and older subgroups but was more pronounced in patients whose initial diagnosis was symptomatic myocardial infarction rather than silent myocardial infarction or angina pectoris. This apparent advantage associated with Type A behavior is surprising and needs confirmation, but the results do indicate that patients with CHD and a Type A behavior pattern are not at increased risk for subsequent CHD mortality.