Fire simulation codes are powerful tools for use in risk-informed and performance-based approaches for risk assessment. Following initial work performed in a joint effort between the U.S. Nuclear Regulatory Commission and the Electric Power Research Institute of a verification and validation of five popular fire simulation codes and research performed at the University of Maryland to quantify total code output uncertainty following a “black-box” approach, this research presents a “white-box” methodology with the goal of also accounting for uncertainties within a simulation code prediction. In this paper the white-box probabilistic approach is discussed to assess uncertainties associated with fire simulation codes. Uncertainties associated with the input variables to the codes as well as the uncertainties associated with the submodels and correlations used inside the code are accounted for. To validate code output calculations, experimental tests may also be available to compare against code calculations. These experimental results may also be used in the assessment of the code uncertainties. Building upon earlier research on model uncertainty performed at the University of Maryland, the methodology employed to estimate the uncertainties is based on a Bayesian estimation approach. This Bayesian estimation approach integrates all evidence available to arrive at an estimate of the uncertainties associated with a reality of interest being estimated by the simulation code. Examples of applications with results of the associated uncertainties are discussed in this paper.