The prognostic value of a left ventricular (LV) mass index (g/m2) estimated from an electrocardiographic model and radiographic estimates of the relative heart volume (ml/m2) and cardiothoracic ratio for predicting the risk of cardiovascular disease mortality were investigated using Cox regression analysis to adjust for age, systolic blood pressure and history of heart attack in 1,807 men (1,609 white, 198 black) and 2,143 women (1,884 white, 259 black). The study population (ages 35 to 74 years at baseline) was followed from 5 to 12 years (average 9.5 years) for cardiovascular disease mortality. LV mass index and relative heart volume were independent predictors of cardiovascular disease mortality among white men. All 3 cardiac size estimates were independent predictors for cardiovascular disease mortality among white and black women. When LV mass index was used as a dichotomized variable to indicate the presence or absence of LV hypertrophy, the age-adjusted relative risk of cardiovascular disease mortality was 2.48 (95% confidence interval 1.77 to 3.46) for white men, 3.03 (1.49 to 6.16) for black men, 1.86 (1.21 to 2.87) for white women and 2.05 (0.83 to 5.05) for black women. The corresponding prevalence of LV hypertrophy was 15.4% for white men, 36.6% for black men, 20.1% for white women and 17.4% for black women. It is concluded that the electrocardiographic estimate of LV mass index can identify a substantially larger fraction of persons at increased risk for cardiovascular mortality than conventional electrocardiographic criteria for LV hypertrophy and that LV mass index estimated by electrocardiogram is a valuable supplement to radiographic cardiac size estimates in epidemiologic applications.