Through performing hydrogen desorption experiments at different heating rates, beta, the (effective) activation energy, E, of the desorption process can be determined from the shift of a characteristic temperature, T-f, of the hydrogen evolution effect with heating rate. In the literature various methods have been employed, and in the present work the accuracy of these methods is investigated. The Kissinger-Akahira-Sunose, Flynn-Wall-Ozawa, Starink, Kissinger and Choo-Lee methods all employ approximations which cause deviations in the activation energy determination, which increase drastically as E/RT (R is the gas constant) becomes smaller. It is shown that for various hydrogen desorption reactions reported in the literature, deviations in reported E between similar to 1 and similar to 20% can occur due to inappropriate use of methods. It is shown that the Ozawa and Flynn-Wall-Ozawa methods are highly inaccurate and particularly for hydrogen evolution, where E/RT is often smaller than 15, they are in most cases inappropriate. The Kissinger peak method is accurate for first order reactions, but as hydrogen evolution reactions generally are not first order reactions, application of this method will result in inaccuracies which increase for decreasing E/RT. In general the magnitude of the deviations of such a peak method are not predictable, as this depends on the reaction mechanism. In many cases the Kissinger peak method is inappropriate for high accuracy determination of activation energy for hydrogen evolution reactions. Amongst the methods that provide an activation energy directly from a slope (i.e. without iterative procedures) the Starink method provides the best accuracy of activation energy analysis methods studied in the literature. It provides an accuracy that is better than 2% for E/RT > 6, which covers all known hydrogen desorption reactions, whilst correction for residual errors are possible. (C) 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.