Patient referrals from general practice for suspected heart failure are increasing the demand for echocardiograms, many of which are normal. We investigated whether general practitioners could be more selective by referring only patients with abnormal electrocardiograms for echocardiography. The electrocardiograms of 200 patients attending a heart failure clinic were analysed by a consultant cardiologist and two general practitioners. All three assessors examined the electrocardiograms independently and unaware of the echocardiography results. The correlation between abnormal electrocardiograms and left ventricular systolic dysfunction on echocardiography was assessed, together with the concordance between the assessors in their electrocardiogram interpretations. One hundred and sixty-five patients had echocardiographic evidence of left ventricular systolic dysfunction. When interpreted by a cardiologist, the electrocardiogram had a sensitivity of 89.1% and a specificity of 45.7% in predicting left ventricular systolic dysfunction. The general practitioners' results were comparable to the cardiologist's. We estimate that using the electrocardiogram to select patients could reduce the number of open access echocardiograms performed for suspected heart failure by up to 43% but would miss 10% of those with significant left ventricular systolic dysfunction. We therefore do not recommend selecting patients for open access echocardiography on the basis of electrocardiographic abnormalities.