Large-scale simulations of multidimensional unsteady reacting flows with detailed chemistry and transport can be computationally extremely intensive even on distributed computing architectures. With the development of computationally efficient reduced chemical kinetic models, the smaller number of scalar variables to be integrated can lead to a significant reduction in the computational time required for the simulation with limited loss of accuracy in the results. A general MATLAB-based automated procedure for the development of reduced reaction models is presented. Based on the application of the quasi-steady-state (QSS) approximation for certain chemical species and on the elimination of selected fast elementary reactions, any complex starting reaction mechanism (detailed or skeletal) can be reduced with minimal human intervention. A key feature of the reduction procedure is the decoupling of the QSS species appearing in the QSS algebraic relations, enabling the explicit solution of the QSS species concentrations, which are needed for the evaluation of the elementary reaction rates. In contrast, previous approaches mainly relied on an implicit solution, requiring computationally intensive inner iterations. The automated procedure is first tested with the generation of an implicit 5-step reduced reaction model for CH4/air flame propagation. Next, two explicit robust reduced reaction models based on ignition data (18-step) and on flame propagation data (15-step) are systematically developed and extensively validated for ignition delay time, flame propagation, and extinction predictions of C2H4/air, CH4/air, and H-2/air systems over a wide range of equivalence ratios, initial temperatures, pressures, and strain rates. In order to assess the computational advantages of the explicit reduced reaction models, comparisons of the computational time required to evaluate the chemical source terms as well as for the integration of unsteady nonpremixed flames for each model are also presented. (c) 2007 The Combustion Institute. Published by Elsevier Inc. All rights reserved.