This paper describes a general methodology for quantifying the importance of specific phenomenological elements to analysis measures evaluated from nonparametric best-estimate plus uncertainty evaluation methodologies. The principal objective of an importance analysis is to reveal those uncertainty contributors having the greatest influence on key analysis measures. This characterization supports the credibility of the uncertainty analysis, the applicability of the analytical tools, and even the generic evaluation methodology through the validation of the engineering judgments that guided the evaluation methodology development. A demonstration of the importance analysis is provided using data from a sample problem considered in the development of AREVA's realistic large-break loss-of-coolant (LOCA) methodology. The results are presented against the original large-break LOCA phenomena identification and ranking table developed by the technical program group responsible for authoring the code scaling, applicability, and uncertainty methodology.