Genetic algorithms, developed by J. H. Holland, are randomized search algorithms which mimic the mechanics of natural selection and natural genetics. In this work a genetic algorithm was applied to optimization of sophisticated thin-film systems. It was found to be an efficient tool rapidly converging to the target as defined by the merit function. The algorithm is introduced, discussed, and illustrated for two applications: a polarization-preserving coating of a penta-roof prism for a long-grange optical sight and a phase light splitter--beam divider/combiner for bi-directional fringe-counting interferometers.