Aim To examine the association of radiographic measures of heart size with mortality from coronary heart disease.

Methods and Results One thousand, one hundred and ninety-one male civil servants aged 40-69 years were followed-up for mortality over 25 years in relation to cardiothoracic ratio and relative heart volume. A high cardiothoracic ratio and relative heart volume predicted coronary (n= 196 deaths) and all-cause mortality, but not respiratory or malignant mortality. After adjustment for age, systolic and diastolic blood pressure, the highest (≥0·47) compared to the lowest quintile of the cardiothoracic ratio (<0·40) was associated with a rate ratio of 1·84 (95% CI 1·14–2·97) for the effect on coronary heart disease mortality. Further adjustment for heart rate, smoking, cholesterol, angina and ECG ischaemia had little effect, reducing the rate ratio to 1·65 (95% CI 1·01–2·70). Similar rate ratios were observed for relative heart volume.

Conclusions Cardiothoracic ratio within the range considered ‘normal’ in clinical practice predicted coronary heart disease mortality independent of established coronary heart disease risk factors. The relative heart volume, which uses measurements from the lateral as well as the posteroanterior chest X-ray, did not predict coronary heart disease any better than the cardiothoracic ratio. The extent to which left ventricular mass and systolic dysfunctionpathophysiological correlates of the cardiothoracic ratio and relative heart volume - are independent risk factors for coronary heart disease should be further investigated.