Exacerbation of heart failure may increase susceptibility to arrhythmias. Therefore tests to assess the risk of arrhythmia, performed after hemodynamic improvement, may be of limited value. To determine whether hemodynamic improvement alters ventricular late potentials detected by signal-averaged ECG, we studied 27 consecutive patients with dilated heart failure (left ventricular ejection fraction 0.20 +/- 0.06, 15 with coronary artery disease) before and 3 +/- 2 days after tailored vasodilator and diuretic therapy reduced ventricular filling pressures. QRS duration, terminal QRS amplitude (root mean square [RMS]), and low-amplitude (less than 40 microV) signal (LAS) duration were determined by an automated algorithm from the vector magnitude of the QRS high-pass filtered at 25 Hz and at 40 Hz. Despite marked decreases in pulmonary capillary wedge (27 +/- 7 to 16 +/- 5 mm Hg, p less than 0.001) and right atrial (13 +/- 7 to 7 +/- 4 mm Hg, p less than 0.001) pressures and a 20% increase in cardiac output, there was not a significant change in QRS duration, RMS, or LAS. Before and after therapy late potentials, defined as abnormal QRS duration RMS, or LAS, were present in 14 (52%) patients with filtering at 25 Hz and in 22 (81%) patients with filtering at 40 Hz. The signal-averaged ECG after hemodynamic improvement predicted the results during exacerbation of heart failure in all patients. Thus in patients with advanced heart failure the signal-averaged ECG obtained after hemodynamic improvement reflects the findings during exacerbation of heart failure.