in numerical simulation of combustion models, solution of the chemical kinetics is often the most expensive part of the calculation, since accurate description of kinetic mechanism involves large number of species and reactions, leading to a large set of coupled ODEs, often too complex to be considered in their entirety along with a detailed flow simulation. Hence the need for representing the complex chemical reactions by simple reduced models, which can retain considerable accuracy while rendering computational feasibility. Realistically, under different conditions and at different points in time, different reactions become important, which has been exploited to develop an adaptive reduction scheme such that the reduced reaction model adapts itself to the changing reactor conditions. A methodology is developed in this paper to construct reduced mechanisms by solving an optimization problem, where the objective is to determine the range of conditions along the reaction trajectory over which a prespecified number of reactions can predict the actual profile within an allowable tolerance. Such an adaptive reduced mechanism is then coupled with the reactive flow algorithm, which selects an appropriate mechanism depending on reactor condition and integrate the corresponding ODEs for the specified valid range. These ideas are demonstrated using the mechanism of CO/H-2 combustion in air. (C) 2003 Elsevier Ltd. All rights reserved.