An improvement of the measurements reliability in thermal processes by the application of the advanced data validation and reconciliation (DVR) methodology strictly depends on their assumed uncertainties. In thermal processes, which are realized in industry, the measurements uncertainties are often determined by the use of all available information concerning possible variability of measurement results. Very often, the determination of measurements uncertainties require professional knowledge, however this approach may lead to imprecise results The paper presents an application of the advanced data reconciliation method with the use of uncertainties of measurements expressed by fuzzy numbers. The proposed methodology has been presented based on example of gas-and-steam CHP unit. In order to compare calculation results for deterministic and the fuzzy uncertainties, the complex standard uncertainties of main assessment indicators of analyzed CHP unit operation supervision as well as the relative entropy of information – Kullback-Leibler divergence for two multivariate normal distributions have been chosen. By the use of mentioned criterions, it has been observed that better results are obtained with the proposed method using fuzzy measurements uncertainties. The results of this study showed that calculated values of uncertainties of energy assessment indicators and the relative information entropy of a whole measurements system are both smaller and more reliable.