Forty-two patients with an ejection fraction of 0.20 or less underwent coronary artery bypass grafting from 1986 to 1990 using a method of myocardial protection we term "centigrade cardioplegia," combining single-dose, cold, crystalloid cardioplegia, systemic hypothermia, and local hypothermia. Thirty-day mortality was 4.8% (2/42). Perioperative morbidity included two myocardial infarctions (4.8%) and one stroke (2.4%), which fully resolved. Postoperative left ventricular function improved (left ventricular ejection fraction, 0.157 +/- 0.028 to 0.226 +/- 0.085; p < 0.0002), as did New York Heart Association class (3.4 +/- 0.73 to 1.8 +/- 0.63; p < 0.0001) and Canadian class (3.3 +/- 0.81 to 0.61 +/- 0.92). Survival, 88% at 1 year, declined to 68% at 3 years and 34% at 6 years. This high-risk group had very acceptable short-term results, indicating adequate intraoperative myocardial protection. Four clinical variables were associated with long-term survival: (1) chief complaint of pain only (p = 0.05), (2) history of unstable angina (p = 0.04), (3) Canadian class less than IV (p = 0.05), and (4) New York Heart Association class less than IV (p = 0.05). Reduced survival, although not statistically significant (p = 0.07), was noted for right ventricular ejection fraction of 0.30 or less. These factors may help predict which patients with severe left ventricular dysfunction will benefit from revascularization.