The present work investigates the accuracy of laminar flame speeds measured from outwardly propagating spherical flames. We focus on methane/air mixtures at normal temperature and pressure, for which there is a variety of data sets reported in the literature. It is observed that there are large discrepancies in laminar flame speed measurement, which makes these experimental data unhelpful for restraining the uncertainty of chemical models. Different sources of uncertainty/inaccuracy (including mixture preparation, ignition, buoyancy, instability, confinement, radiation, nonlinear stretch behavior, and extrapolation) are discussed and their contributions to large discrepancies in laminar flame speed measurement are assessed with the help of 1-D simulation. It is found that the uncertainty in equivalence ratio can bring large inconsistency in laminar flame speed measurement, especially for off-stoichiometric mixtures and experiments using pressure gauge with normal or low accuracy. For fuel-rich methane/air mixtures, the large deviations in laminar flame speed measurement could be partly caused by nonlinear stretch behavior and extrapolation. The change of the influence of different sources of uncertainty with initial pressure, initial temperature, and fuel carbon number is also discussed. Furthermore, it is shown that the discrepancy in raw experimental data can be possibly hidden after extrapolation is conducted. Therefore, the data used for extrapolation as well as extracted results should be reported and compared with simulation or other experiments. The recommendations on the laminar flame speeds measurement using the propagating spherical flames are also provided. (C) 2015 The Combustion Institute. Published by Elsevier Inc. All rights reserved.