Two hundred thirty-six patients with peripheral vascular disease were prospectively studied to assess whether noninvasive cardiac investigations could predict prognosis better than simple clinical assessment. Clinical history, examination and resting electrocardiography were considered in all patients; exercise electrocardiography, Holter monitoring, radionuclide ventriculography and dipyridamole thallium imaging were performed in a subgroup of 168 patients. Follow-up for 6 to 30 months revealed major cardiac events in 21 patients. Cox survival analysis showed that clinical evidence of prior coronary artery disease was the best variable from clinical assessment that predicted cardiac events, with no other clinical variable adding to the statistical model. When variables from noninvasive cardiac assessment were added to the model, which included clinical evidence of coronary artery disease, dipyridamole thallium heart:lung ratio and left ventricular ejection fraction added significantly and incrementally to the prediction of cardiac events. Results of exercise electrocardiography or Holter monitoring did not add significantly. It is concluded that high lung uptake of thallium during dipyridamole stress, and impaired left ventricular ejection fraction help to identify patients with peripheral vascular disease who are at high cardiac risk, and should therefore be used for selecting subsequent cardiovascular medical, surgical and anesthetic management.