This report presents the 42-month survival experience of 91 patients with a significant lesion of the left main coronary artery in the Veterans Administration Cooperative Study of Coronary Bypass Surgery. Survival in surgical patients was significantly better than that in the medical group (p = 0.016), even after adjustments were made for two important differences in baseline characteristics--duration of angina and high risk by angiographic criteria--between the two groups (p = 0.019). Subgroups based on severity of left main stenosis and on left ventricular (LV) function showed significant trends in favor of surgery in patients with more than 75% left main stenosis and in those with abnormal LV function. A similar but nonsignificant trend was seen in the two subgroups with 50-75% stenosis or with normal LV function. The surgical benefits were not significantly different between the categories of the subgroups defined separately by stenosis and LV function. Low-, middle- and high-risk subgroups based on four noninvasive clinical predictors also showed significantly improved survival with surgery in the high-risk group. The low-risk groups showed a slight, nonsignificant disadvantage with surgical treatment. These data support the view that patients with left main disease are not a homogeneous group. High- and low-risk subgroups with different outcomes and responses to treatment can be delineated by angiographic or clinical criteria. For most patients with left main disease, coronary artery bypass grafting offers improved longevity.