BACKGROUND We determined that cold blood cardioplegia and intermittent ventricular fibrillation with ischemia were associated with similar enzyme and myocardial protein leakage in a randomized, prospective study of 40 patients. We have continued to use both methods in our unit, according to surgeons' preference. METHODS In our database we have reviewed 1,923 patients who have undergone first-time elective or urgent coronary artery bypass grafting from January 1992 to May 1997. RESULTS Five hundred seventy-eight patients underwent coronary artery bypass grafting with cold blood cardioplegia and 1,345 had ventricular fibrillation and aortic cross-clamping. The preoperative factors were virtually identical. Intraoperative differences were only those inherent to the two techniques: temperature and cross-clamp time. Mortality was 2.5% for ventricular fibrillation and aortic cross-clamping arrest and 2.1% for cardioplegia (p=0.55 by chi2 test). There was a higher use of the intraaortic balloon pump in the ventricular fibrillation and aortic cross-clamping group (2.4% versus 1.0%; p=0.04), but no other differences in outcome were detected. CONCLUSIONS A truly randomized trial to demonstrate which technique is superior is impractical at this level of difference because it would require 37,000 patients to avoid a beta error. We have to base our practice on the retrospective data available. Each technique has its merits in practice, which are discussed.