A variety of numerical simulation tools and computer codes are widely used in the nuclear safety community to evaluate the status of the nuclear power plants during accident conditions. Currently, for this purpose the “best-estimate approach”—covering best-estimate initial conditions, use of best-estimate code, and uncertainty and sensitivity analysis—is used. So far, the quantification of uncertainties was not widely applied for the analysis of severe accidents due to the complexity of the processes taking place at the different phases of severe accidents and possible inaccuracies of numerical tools and computer codes. This work presents the experience of Lithuanian Energy Institute experts in severe accidents analysis, summarized in the frame of the HORIZON 2020 EURATOM project “MUSA.” The modeling of PHEBUS FPT1 experiment, severe accidents in boiling water reactor and in spent fuel pool were performed using codes for severe accident analysis RELAP/SCDAPSIM and ASTEC. The focus was on identifying uncertainty trends and sensitivity analysis. The performed uncertainty and sensitivity analysis revealed the trend indicating uncertain parameters that most influenced the results of severe accident calculation within different phases of accident. Knowing which uncertain parameters have the greatest influence on the calculation result, special attention must be paid to determining these values. This would reduce the width of uncertainty bands, which allows for more accurate selection of accident management measures, which in turn will improve nuclear safety.