Laminar flame speed has traditionally been used for the partial validation of flame kinetics. In most cases, however, its accurate determination requires extensive data processing and/or extrapolations, thus rendering the measurement of this fundamental flame property indirect. Additionally, the presence of flame front instabilities does not conform to the definition of laminar flame speed. This is the case for Le < I flames, with the most notable example being ultralean H-2/air flames, which develop cellular structures at low strain rates so that determination of laminar flame speeds for such mixtures is not possible. Thus, this low-temperature regime of H-2 oxidation has not been validated systematically in flames. In the present investigation, an alternative/supplemental approach is proposed that includes the experimental determination of extinction strain rates for these flames, and these rates are compared with the predictions of direct numerical simulations. This approach is meaningful for two reasons: (1) Extinction strain rates can be measured directly, as opposed to laminar flame speeds, and (2) while the unstretched lean H2/air flames are cellular, the stretched ones are not, thus making comparisons between experiment and simulations meaningful. Such comparisons revealed serious discrepancies between experiments and simulations for ultralean H-2/air flames by using four kinetic mechanisms. Additional studies were conducted for lean and near-stoichiornetric H-2/air flames diluted with various amounts of N-2. Similarly to the ultralean flames, significant discrepancies between experimental and predicted extinction strain rates were also found. To identify the possible sources of such discrepancies, the effect of uncertainties on the diffusion coefficients was assessed and an improved treatment of diffusion coefficients was advanced and implemented. Under the conditions considered in this study, the sensitivity of diffusion coefficients to the extinction response was found to be significant and, for certain species, greater than that of the kinetic rate constants. (c) 2005 The Combustion Institute. Published by Elsevier Inc. All rights reserved.