The paper addresses the reproducibility of the fire test in the United Nations "Global technical regulation on hydrogen and fuel cell vehicles" (GTR#13) and similar fire test protocols in other regulations, codes and standards (RCS). Currently, GTR#13 requires controlling the flame temperature beneath the tank. An original Ulster conjugate heat transfer numerical model was applied to carry out a study demonstrating the dependence of a fire resistance rating (FRR) of a composite hydrogen tank on a fire heat release rate (HRR). No thermally activated pressure relief device was used. The validation experiments conducted afterwards at Karlsruhe Institute of Technology (KIT) plus a former USA fire test have confirmed the Ulster's conclusion to control not only temperatures, yet the fire HRR. This will improve the GTR#13 fire test reproducibility in different laboratories worldwide. The numerically observed variations of FRR were confirmed by the unique experimental data of the authors' collaborators: FRR = 16-22 min for HRR = 79 kW, 7-8 min (HRR = 165 kW) - both tests were carried out at KIT with identical 36 L volume and 700 bar pressure tanks; and 6-7 min (HRR = 370 kW), though this test in USA was performed with a larger volume tank of 72.4 L and 350 bar. The data on pool fire test with significantly higher HRR, i.e. 4100 kW, and tank volume of 100 L and 700 bar pressure confirmed the "saturation" effect in the dependence of FRR on HRR at HRR above 350 kW. The results of the study underpin the suggested amendment to GTR#13 to improve the reproducibility of the fire test and perform tests with onboard storage tanks at HRR>350 kW. (C) 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.