The sampling literature contains many examples of estimators of population parameters. In the case of generalization of these estimators, estimation of optimum values is essential. For the optimum estimator, the mean square error equations are minimized with respect to unknown parameters and nonlinear constraints. In this paper an attempt is made to get these optimum estimators of parameters in stratified random sampling using genetic algorithms (GAs). Numerical examples are given to illustrate the algorithms. The results show that the genetic algorithm is more efficient than classical ratio type estimator in stratified random sampling.