Modular redundancy is more effective than component redundancy, as a modular scheme of redundancy allocation in multilevel systems not only enhances the system reliability but also provides fault tolerance to the optimum design. Therefore, to increase the efficiency, reliability and maintainability of a system, the modular redundancy should be considered instead of traditional approach of component redundancy. Multi-level redundancy allocation problem in fuzzy environment has been formulated and solved using advanced genetic algorithm (GA) and penalty function technique. Numerical examples have been solved and the results have been discussed.