Optimal management of coarctation with ventricular septal defect is difficult. Should one treat the coarctation, the ventricular septal defect, or both? This dilemma was investigated by reviewing 39 successive patients, aged less than 4 months, undergoing coarctation repair without pulmonary artery banding. Twelve hospital deaths occurred. Ventricular septal defect size was graded "blind" according to preoperative angiocardiographic and haemodynamic findings. Of 14 patients classified as having a large ventricular septal defect, necropsy and operative findings in eight showed defects 7 to 12 mm in diameter. Factors associated univariately with significantly increased mortality were young age, raised atrial and ventricular end-diastolic pressures, low weight, high admission blood urea, preoperative ventilation, and a large ventricular septal defect. All but the last two were also closely associated with each other. A jack-knifed discriminant function based upon ventricular septal defect size, blood urea, pulmonary venous oxygen content, and inferior caval oxygen saturation correctly predicted outcome in 78.9% of patients. Combinations of these four giving a probability greater than 0.9 of survival were rare with blood ureas above 8 mmol/litre. Raised blood urea was associated with low descending aortic pressure and subsequent dialysis. Only patients with a large ventricular septal defect stand to benefit from pulmonary artery banding at initial operation. Preoperative treatment to increase renal blood flow (prostaglandins, dopamine) may improve overall survival.