The influence of ECG abnormalities on the incidence of sudden cardiac death is examined in the Framingham Study cohort and is compared with the effect of known cardiovascular risk factors. Although many individual ECG abnormalities are associated with SD, multiple regression analysis, including non-ECG variables, leaves fewer independently related ECG sudden death risk factors: ECG-MI and IVB in those with overt CHD, LVH and tachycardia in all men and in women without CHD, and nonspecific ST-T abnormalities only in men. VPBs appear to be without an independent influence. Inclusion of ECG variables in the predictive model for SD has little effect on identifying those at risk for sudden unexpected death, but it improves substantially the predictive value in persons with known CHD.