BACKGROUND: In a prospective epidemiological study of chronic Chagas' disease, several clinical and echocardiographic variables were analyzed as predictors of mortality. METHODS: Among 960 subjects seropositive for Chagas' disease who were examined between June 1981 and June 1992, 283 had echocardiograms. RESULTS: During a mean follow-up period of 48.3 +/- 36.4 months (range, 1-156 months), 108 subjects died. Echocardiographic end-diastolic and -systolic left ventricular internal dimensions, fractional shortening, radius-to-thickness ratio, left ventricular mass, mitral E-point septal separation, and 17 other nonechocardiographic variables were predictors of death on univariate analysis (P < 0.001 for each). On stepwise multiple regression analysis of 215 subjects, significant risk covariates in a Cox model analysis were clinical group (P < 0.0001), M-mode echocardiographic E-point septal separation of 22 mm (P = 0.003), presence of first- or second-degree heart block (P = 0.003), chest radiologic cardiothoracic ratio >/= 0.55 (P = 0.012), presence of electrocardiographic ST segment elevation on precordial leads (P = 0.014), age >/= 56 years (P = 0.028), and presence of right bundle-branch block (P = 0.045). Patients with an apical aneurysm on two-dimensional echocardiography had an increased mortality (Chi-square = 11.5, P < 0.001). CONCLUSIONS: Echocardiography is a valuable tool to assess the risk of death in prospective studies on chronic Chagas' heart disease.