An application of the Best Estimate Plus Uncertainty (BEPU) method is made to an analysis of the “Intentional depressurizaion of steam generator secondary side” which is an accident management procedure in a small-break loss-of-coolant accident (SBLOCA) with high pressure injection (HPI) system failure. RELAP5/MOD3.2 is used as the system analysis code. Interfacial friction in the core affects the two-phase mixture level and the distribution of the dispersed gas phase. This phenomenon is very important in terms of the influence its uncertainty has on the peak cladding temperature. The RELAP5/MOD3.2 code uses drift-velocity to describe the interfacial friction coefficients in vertical dispersed flow. The Chexal-Lellouche drift-flux correlation is used for the rod bundle geometry. In the present study, the RELAP5 model uncertainty was quantified regarding the interfacial friction coefficients in the rod bundle geometry by conducting numerical analyses of separate effect tests. As the separate effect tests, two-phase mixture level swell tests in the Thermal Hydraulic Test Facility (THTF) of the Oak Ridge National Laboratory (ORNL) were used. After considering applicability to the SBLOCA, tests were selected for which conditions of pressures and rod powers were similar to PWR plant conditions. A total of 55 data were used. The model uncertainty parameter was defined as a multiplier for the interfacial friction coefficient. Numerical analyses were performed by adjusting the multiplier so that the predicted void fractions agreed with the experimental measured data. The resultant distribution of the multipliers represented the model uncertainty. The mean, standard deviation, minimum and maximum values of this uncertainty distribution were 0.88, 0.55, 0.13 and 3.0, respectively.