Uncertainty analysis methodology in complex system models requires a comprehensive treatment of many different types and sources of uncertainty sources. Proposed methodology as part of a broader research by the authors for an integrated methodology for uncertainty analysis of TH computational codes, allows updating of uncertainties upon availability of new information about inputs, models or output. This methodology is a statistical approach for output-based updating, with non-paired data. The proposed method employs a Gibbs MCMC sampler to estimate output distribution. Winbugs code was developed to implement the Bayesian solution proposed in this paper. The methodology is applied to the data from 200% cold leg LBLOCA test of LOFT facility [4].