Several researchers have endeavored to develop methodologies to extrapolate the uncertainties gathered from reduced-size facilities to the full-size nuclear power plant. They are all based on the general guideline of the code scaling, applicability, and uncertainty (CSAU) method. Although there is an extensive compilation of experimental and theoretical databases and a detailed guide about the best-estimate calculation of loss-of-coolant accidents, these applications are dissimilar to each other. The absence of a procedure to implement the requirement of direct data comparison with integral- and separate-effects tests in determining the code uncertainty is the main cause of the differences. To overcome this problem, a code-accuracy-based uncertainty estimation (CABUE) technique has been developed, in which the code accuracy becomes the measure of the selection of code parameters and the determination of the ranges of them. An application of this technique to a Westinghouse three-loop nuclear power plant has been successfully performed.