Consistent datasets of experiments are highly important both for validation and optimization of kinetic mechanisms. An analysis of the data consistency of all available burning velocity measurements of hydrogen flames using the heat flux method at atmospheric pressure is performed in the present work. A comparison of many experiments performed in several laboratories with different types of dilution by various inerts was guided by kinetic modeling using two kinetic mechanisms. Konnov (2015) and ELTE (Varga et al., 2016) models demonstrated a uniform trend at all conditions tested: the second mechanism predicts lower burning velocities which are in better agreement with the heat flux measurements from different groups. Some experimental datasets, however, significantly disagree with one or both models; these conditions were revisited experimentally in the present work. The laminar burning velocities of H-2 + O-2 + N-2 mixtures with 7.7% O-2 in O-2 + N-2 oxidizer and of 85:15 (H-2+ N-2) and 25:75 (H-2+ N-2) fuel mixtures with 12.5:87.5 (O-2 + He) oxidizer have been measured. It was concluded that the results of Hermanns et al. (2007) are somewhat higher than those of other studies at similar conditions and a possible reason of this disagreement was suggested. Analysis of the measurements performed by Goswami et al. (2015) on a high-pressure installation suggests an equipment malfunction that led to the erroneous values of the equivalence ratio for hydrogen and syngas flames. The ELTE mechanism developed using an optimization approach shows a very good performance in predicting laminar burning velocities of hydrogen flames measured using the heat flux method. (C) 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.