A prognostic index for predicting significant coronary artery disease was established using multiple logistic regression analysis of clinical data from 643 patients with valvular heart disease who had undergone routine coronary arteriography before valve replacement. The index or equation obtained incorporated the presence of angina, a family history of ischaemic heart disease, age, cigarette smoking habits, mitral valve disease, sex, and electrocardiographic evidence of myocardial infarction. The equation was validated using prospective data from 387 patients with valvular disease and shown to enable almost a third of routine coronary arteriograms to be omitted while maintaining 95% sensitivity for patients with coronary artery disease. Similar analysis of the more detailed prospective data produced a second discriminant function incorporating diastolic blood pressure, total cigarettes smoked in life, the severity of angina, family history of ischaemic heart disease, age, current cigarette smoking habits, and the ratio of total to high density lipoprotein cholesterol. This method improved the discrimination between patients with and without coronary artery disease, allowing omission of 30% of routine coronary arteriograms with 100% sensitivity for patients with coronary disease and omission of 41% with a 96% sensitivity level.