1. An open-circuit method for the measurement of oxygen consumption in children which is suitable for use during cardiac catheterization has been described. 2. Since it has been previously shown that there is no systematic error in this method, the accuracy of the method is described by the random error (i.e. reproducibility). The reproducibility of the method has been determined by the repeated measurement of oxygen consumption in children with congenital diseases of the heart under specified ‘standard’ conditions. 3. A simple method of predicting the oxygen consumption in these children was devised by subjecting the data to a multivariate regression analysis. 4. It is concluded that the accuracy of these predictions, although considerably better than that in the published literature, is not good enough to merit its use in the routine haemodynamic assessment.