A method providing initial estimates for rate constants of nonlinear chemical kinetics is proposed. This method is based on a genetic algorithm. The particular structure of the genetic algorithm adapted to the problem of estimating rate constants is introduced. The rate and certainty of convergence of the estimation procedure is analysed illustrating the proposed strategy by example of its application to a kinetic model of the catalytic oxidative coupling of methane to C-2 hydrocarbons. For this example, the maximum likelihood estimate of the rate constants could be found using initial estimates from the genetic algorithm which were further optimized by a Nelder-Mead algorithm. It is demonstrated that the optimization procedure proposed supports the estimation of rate constants for kinetic models without supposition of rate-determining steps.