A new method for evaluating kinetic flame models, called net rate analysis, was developed from existing mechanism screening techniques and a modified form of reaction path analysis. In net rate analysis, the contributions of individual reactions are gauged by comparing rates of reaction computed using experimental mole fractions, rather than mole fractions produced by a flame code simulation. The sum of these contributions for a given species-the data-based predicted net rate-is compared to the experimental net rate derived from measured mole fractions using the laminar flame equations. Various features of a model＇s structure and its predictions, such as interdependencies of submechanisms and net rates, can be revealed with this technique. Measured profiles of mole fractions and temperature in flat, laminar, premixed H-2/O-2/C6H6/Ar and C6H6/O-2/Ar flames were used to evaluate the new technique against conventional reaction path analysis or mole fraction comparisons. In some cases, the conventional methods appear to identify the existence of a problem with the chemistry of a particular species when, in fact, the chemistry is correct. Propagation of errors in the predicted mole fraction of other species to the species under consideration are the source of the discrepancy. Also observed in the use of the conventional analyses are cases where deficiencies in the chemistry involving a particular species are offset by inaccurate predictions of other species, thus leading to the incorrect conclusion that the proposed chemistry is sound. Net rate analysis is shown to lead to more realistic conclusions than techniques based on the use of computed mole fractions because the net rate analysis eliminates the propagation of error found in predicted mole fractions.