Abstract Warm heart surgery—37°C cardioplegia with systemic normothermia—has been introduced as an alternative to conventional hypothermic cardiac surgery. A randomised trial comparing warm (W) and cold (C) methods was done in 1732 patients undergoing isolated coronary bypass surgery in three adult cardiac surgery centres at the University of Toronto, Canada. Allocation to W (860 patients) or C (872) was stratified by urgent versus elective operations and by surgeon. There were no striking baseline differences in patients' demographics, angiographic findings, or operative procedures. All but 4·2% of patients initially received antegrade cardioplegia; a further 2·1% switched to retrograde delivery intra-operatively. Crossovers to C occurred in 7·7% of cases either due to difficulty in sustaining cardiac arrest or due to coronary flooding. Analysis, however, was on an intention-to-treat basis. The 30-day all-cause mortality was 2·5% in C patients and 1·4% in the W group (p 0·12). There was no difference in non-fatal Q-wave infarction rates (W 10·1%, C 11·1%), but enzymatic infarction by serial creatine kinase MB fraction (CK-MB) measurements was reduced (W 12·3% vs C 17·3%, p<0·001) as was the mean area under the CK-MB curve. Postoperative low-output syndrome was less frequent in W patients (6·1% vs 9·3%, p 0·01). There were no differences in the rates of stroke, reoperation for bleeding or tamponade, or sternal rewiring/ debridement for dehiscence or infection. Warm heart surgery is a safe and effective alternative to conventional hypothermic techniques for patients undergoing coronary bypass surgery.