In order to define the best strategy of prognostic evaluation in relation to patients in heart failure, 415 patients with impaired left ventricular function (ejection fraction < 40%) were prospectively included, between June 1993 and 1996, in a study comparing the respective value of high amplification ECG and Holter analysis of sinus variability, 308 patients in sinus rhythm and narrow QRS complexes were included. The patients were distributed into 4 groups according to the presence or absence of late potentials and altered sinus variability (group I with 2 normal examinations, group II with late potentials and normal variability, group III without late potentials, but with altered variability and group IV with 2 abnormal examinations). The sudden and overall mortality was significantly greater in groups III (28 and 11%) and IV (28 and 9%) than in group I (7 and 3%) and II (11 and 2%). The presence of potential was unable to identify patients at risk of ventricular tachycardia and sudden death. In another 60 patients with complete branch block, the prognosis was also correlated with the alteration of sinus variability. In conclusion, high amplification ECG correctly evaluated the prognosis of subjects in heart failure. However, analysis of sinus variability should be systematically proposed to detect subjects at high risk of mortality, whether the subject has large or narrow QRS complexes.