The objective was to prospectively validate a method of increasing the sensitivity, specificity and negative predictive value of a normal ECG in the exclusion of left ventricular systolic dysfunction by the addition of clinical history. We performed a prospective three year study of all referrals to our direct access ECHO service for assessment of LV function. The ECG was reported blind of the result of the ECHO, history of MI or not was noted, and result of the ECHO predicted. Over three years 416 patients were assessed for the presence or absence of left ventricular systolic dysfunction and consequent changes in clinical management. A total of 320(77%) of patients referred with suspected left ventricular dysfunction were found to have normal left ventricular function. Of the 250(60%) patients treated prior to referral for assessment, 183(73%) were treated inappropriately. The combination of a normal ECG and a negative history of myocardial infarction had a sensitivity of 98% and a negative predictive value of 99% in the assessment of LV function. This was an improvement over a normal ECG alone. Our study shows that diagnosis and treatment of heart failure in the community remains sub-optimal. The combination of a normal ECG and no previous history of myocardial infarction is shown to be a sensitive and accurate predictor of normal left ventricular function. If adopted by general practitioners this would be a valuable method of optimising the use of echocardiography in patients with suspected left ventricular dysfunction.