The prediction of requirement for, and short- and long-term outcome of, prolonged mechanical ventilation after cardiac surgery is ill-defined. The aims of this study were to isolate any predictive indices which might identify those groups of patients who may require prolonged ventilation postoperatively and to determine which factors significantly affect outcome in the prolonged-ventilation group. Following case note review of 139 consecutive cardiac surgical patients ventilated for ⩾7 days following surgery, 43 factors were recorded on each patient, including smoking, pulmonary function, chest infection, and chronic obstructive airways disease. Of 139 patients, 89 were discharged from hospital (64% survival); of these, 52 were alive at long-term follow-up (58% long-term survival). Statistical analysis identified urban residence, chronic obstructive airways disease, prolonged operation, and bypass time as significant predictors of requirement for prolonged ventilation postoperatively. On multivariate analysis five factors were predictive of increased intensive care mortality, including urban residence, inotrope days, sepsis, perioperative cerebrovascular accident and coagulopathy requiring fresh frozen plasma transfusion postoperatively. Following discharge from hospital, four factors were found to be significant predictors of increased mortality: these are impaired preoperative ejection fraction, increasing age, impaired preoperative pulmonary function, and abscence of pre-operative aspirin medication. These factors should be considered in intensive care planning, long-term follow-up and importantly on clinical decision making in the individual patient.