Twenty-four patients who had undegone mitral valve surgery for pure non-rheumatic mitral regurgitation were studied non-invasively six months to six years postoperatively. The long-term results of operation were assessed on the basis of clinical history, echocardiography, and treadmill stress testing using a points scoring system. The score so obtained was used to divide the patients into those with a good response to surgery (group 1) and those responding poorly (group 2). The effects on the long-term surgical outcome of several intraoperative and preoperative factors were then analysed both together and separately. A short symptomatic history (less than 1 year), a normal left ventricular end-diastolic volume index (less than or equal to 100 ml per m2), and a large post-ectopic potentiation of KV max (greater than 50 s-1) were found to be favourable prognostic factors when analysed independently. An angiographic ejection fraction less than 0.5 was uniformly associated with a poor outcome, and 71 per cent of patients in atrial fibrillation at the time of operation also responded badly. In those patients with good long-term function, cold potassium cardioplegia was more commonly used than intermittent aortic cross clamping as the means of intraoperative myocardial preservation, though this difference did not reach conventional significance. A standard analysis of variance allowed assessment of length of history, left ventricular end-diastolic volume index, and type of valve prosthesis simultaneously. This indicated that both length of history and left ventricular end-diastolic volume index were highly significant prognostic factors. The use of a Björk-Shiley as opposed to a Starr-Edwards prosthesis also emerged as significantly favouring a good long-term result. The state of the left ventricular myocardium before operation and the type of valve prosthesis used were thus shown to be the prime determinants of surgical outcome in these patients. The optimum time for operation was shown to be within one year of the onset of symptoms, and before the left left ventricular end-diastolic volume index exceeds 100 ml per m2, or the ejection fraction falls to less than 0.5. At such a time, irreversible changes in myocardial function sufficient to negate the beneficial effects of mitral valve surgery have not yet occurred.