Improvements in thermal margin analyses have, in recent years, been dominated by the use of statistical methods to combine the correlation uncertainty with other key uncertainties such as those in temperature, power, or the radial power factor. The methods of combination have included both root-sum-square and Monte Carlo. The latter method provides insight into the probability density function of the composite departure from nucleate boiling ratio, which, for a measured-to-predicted-normal correlation uncertainty and commonly used parameter uncertainties, proves to be skewed slightly upward from normal. In such a case, the assumption of normality is a conservative assumption in licensing analyses.