[en] In a recent international meeting (SNL, MCAP-2003) which has introduced current MELCOR (severe accident analysis computer code) development activities, existing MELCOR core release models failed to predict ISP46 test (international fuel behavior experiments) results well and the latest ORNL-Booth model was suggested to be an optimum model. The need for this kind of model update has been proposed also in the domestic FPT-1 test analysis. Therefore, this research analyzes releases of selected representative volatile and non-volatile radionuclides during conservative high and low pressure sequences in the APR1400 plant using three core release models (CORSOR, CORSOR-M, CORSOR-Booth) in the latest MELCOR 1.8.5 version and the ORNL-Booth update model. As a research strategy, the difference in release fractions among existing and update models is compared and the uncertainty range is then evaluated. The MELCOR fission product core release calculations, which are based on the CORSOR models developed by Battelle Memorial Institute, are performed using various CORSOR empirical release correlations. These correlations assume that fission products are released from the fuel-cladding gap when a failure temperature criterion exceeds or intact geometry is lost and their release rates are based on fuel temperatures. In the analysis, the option of the fuel component surface-to-volume ratio in the CORSOR and CORSOR-M models and the option of the high and low burn-up in the CORSOR-Booth model are considered together. In addition, the update model simulates the effect of Molybdate (Cs₂MoO₄) compound, which is known to be created from the combination of Cs and Mo species during core release, by modifying vapor pressure data. As the results, the CORSOR/CORSOR-M release rate is high for volatile radionuclides, and the CORSOR release rate is high for non-volatile radionuclides with insufficient consistency. As the uncertainty range for the release rate expands from several times (volatile radionuclides) to more than maximum 10,000 times (non-volatile radionuclides), user's careful choice for the core release models is needed. In the trend, the updated ORNL-Booth model is similar with the CORSOR-M model for volatile and mid-volatile radionuclides while it is similar with the CORSOR-Booth model for non-volatile radionuclides. From these, the ORNL-Booth model is recommended for volatile and mid-volatile radionuclides based on the results of the ISP-46 international test and this research, though an optimum model is not fixed until now. On the contrary, no specific model is recommended for non-volatile radionuclides due to insufficient and inconsistent results. This research is the first domestic results evaluating the core release models in the commercial nuclear plants other than in the experiments and is expected to be used as a reference for the source term uncertainty analysis in the domestic nuclear plants