Chronic heart failure (CHF) is associated with high mortality, and there are several established clinical and laboratory parameters that predict mortality in CHF. The purpose of this study was (a) to identify the best ECG parameter that predicts mortality, (b) to evaluate the prognostic marker of ECG against well-established indicators of prognosis. Relevant data from 241 CHF patients were analysed retrospectively. Cardiopulmonary exercise testing and radionuclide ventriculogram were also performed where possible. The mean follow-up period was 31 months. On univariate analysis by the Cox proportional Hazard method, intraventricular conduction delay (IVCD) [P<0.0001, hazard ratio 1.017 (1.011-1.024)] and QTc [P<0.0001, hazard ratio 1.012 (1.006-1.017)] were identified as predictors of mortality. On bivariate analysis, IVCD and MVO2 were better predictors when combined together. A model based on multivariate analysis showed that IVCD, MVO2 and left ventricular ejection fraction (LVEF) were the best predictors of mortality. The addition of plasma sodium, age and NYHA class had no added benefit on the predictive power of the model. Further analysis of IVCD and QTc showed that, for different cut-off values, IVCD is better than QTc, and that there is a graded increase in mortality with increasing value of IVCD. We have found that IVCD is an important ECG predictor of prognosis in patients with CHF.