The relationship between ECG abnormalities and mortality was studied in 4797 males and 4320 females aged 25 to 74 years who took part in the Belgian Inter-university Research on Nutrition and Health (The BIRNH study). At entry all were free of angina, had no history of acute myocardial infarction and showed no evidence of an old infarction on their baseline ECG. They were followed for an average of 5.6 years, and follow-up for vital status was completed satisfactorily in 99.5%. ECG abnormalities were grouped using several classifications: any abnormality, major and minor abnormalities, ischaemic changes, left ventricular hypertrophy and the separate Minnesota codes IV (ST depression), V (abnormal T-wave) and VIII (arrhythmias). Using logistic regression analysis, adjustment of odds ratios for cardiovascular disease (CVD) mortality was done for age, systolic blood pressure, serum total cholesterol and uric acid, diabetes, smoking and antihypertensive drug treatment. Men outnumbered women more than twice in total and CVD mortality. Multivariate analysis showed that the presence of major abnormalities on the ECG was significantly related to CVD mortality in both men and women (adjusted odds ratios 2.73 and 4.40 respectively). In contrast, minor abnormalities were not independently associated with CVD mortality. In men, ST depression (OR = 5.58), signs of an ischaemic ECG (OR = 3.02) and an abnormal T-wave (OR = 2.58) were independently related to CVD mortality. In women primarily a ST depression (OR = 5.87) and arrhythmias (OR = 4.22) had a significant independent effect on CVD mortality.(ABSTRACT TRUNCATED AT 250 WORDS)