In the previously published part 1 of the paper the uncertainty analysis of the large break loss-of-coolant accident (LBLOCA) for German PWR Konvoi was performed using a statistical method, which is based on the Wilks' theory. The evaluated output parameter is the peak cladding temperature (PCT). The primary goal of this (second) part of the paper is a ranking of the input uncertainties, according to their contributions to the PCT uncertainty in the ATHLET simulation of PWR LBLOCA, by performing a sensitivity analysis. It was shown, that the first extended set of varied parameters used in part 1 can be considerably reduced without any statistically significant influence on the uncertainty analysis results. Thus, it can be shown that the input uncertainty vector based on the LBLOCA PIRT of AREVA GmbH was complete. To minimize the number of varied parameters the statistical t-test was used and, thus, a set of uncertainty parameters with significant impact on the uncertainty of the PCT was identified. The main contribution to the uncertainty of the first cladding temperature maximum during the blowdown phase of the accident is produced by the core parameters that affect the fuel's stored energy at the beginning of the accident. However, the major contributors to the uncertainty of the second PCT maximum are the uncertainties in the code models, and first of all the uncertainties in the heat transfer coefficients for dispersed and pure steam flows.