A combination of physical findings, x-ray films and fluoroscopy, electrocardiography, vectorcardiography, cardiac catheterization including oxygen analysis, indicator-dilution curves, and selective angiocardiography was employed to evaluate 259 consecutive patients undergoing right heart catheterization. All patients were classified on the basis of existing shunts. The present report concerns 63 patients with no shunts and 70 patients with a left-to-right shunt through a ventricular septal defect alone.A method employing oxygen analysis of right heart blood was evaluated as an isolated diagnostic technic. Blood specimens were obtained in rapid sequence during withdrawal of the catheter from the pulmonary arteries. Spectrophotometric measurements of blood oxygen saturation were appropriately grouped. The difference in means between the right ventricular outflow tract and right atrium was standardized with the estimated standard deviation and the statistical significance determined from thet-distribution.Reliability was defined as the probability of correctly excluding a shunt when none exists. Sensitivity was defined as the probability of diagnosing an existing shunt. By use of the 1-per cent level of signfiicance as critical the method employed was 98.4 per cent reliable and 97.1 per cent sensitive.Sequential sampling is of vital importance. The value of larger numbers of specimens is illustrated, and the relationship of sensitivity and reliability to the size of the shunt and the number of blood specimens analyzed is illustrated.The reliability and sensitivity of our method and criteria appear superior to previous criteria currently in use.