A genetic algorithm (GA) has been applied to the estimation of parameters appearing in the rate equation of a heterogeneous catalytic reaction. The GA was found to access the global minimum even though the ranges of the parameters were extremely wide and in spite of local minima in the parameter space. The effect of the GA running parameters on the GA performance was studied in detail. For the objective function illustrated in this study low crossover probability with relatively high mutation probability was required for a good performance of GA. Due to the strong dependence of the GA performance on the GA running conditions, a hybrid GA algorithm based on the iteration of the GA running parameters followed by the Levenberg-Marquardt optimizer was developed. The hybrid GA has been found to be efficient and accurate, provided that a proper balance between convergence and diversity was maintained throughout the GA run.