: One of the current problems in optimizing systems reliability field are reliability- redundancy optimization problem. In this thesis, the availability- redundancy optimization approach for series-parallel systems is examined. In recent researches, in order to make the problems of reliability more practical, the performance of system components are considered in three states, healthy, damaged and failed. In these kind of systems, a new definition of Markov process states have been used for evaluating the reliability of series-parallel systems with the same and non-reparable components . In this thesis, the purpose is modeling the reliability- redundancy optimization problem for series-parallel systems with three-state components in following conditions: 1- systems with non-reparable and same components in each subsystem, 2- systems with reparable and same components in each subsystem, 3- systems with non-reparable and dissimilar components in each subsystem and 4- systems with reparable and dissimilar components in each subsystem. The main difference between this study and previous studies is using the markov process with a new definition of states in order to evaluate the availability of series-parallel system under the expressed conditions. The availability- redundancy optimization problem is an NP-Hard problem and new hypotheses make the mathematical model more complex, thus in order to solve the presented models, the genetic metheuristic algorithm is used. The results showes the validity of the proposed models. In this study, the proposed method of Markov process for evaluating system availability is compared with the universal generating function method. The results show that the proposed model provides a similar solution to universal generating function and much lower calculation time. Also, as the system dimensions increase, the difference computing time of the two methods increases. So in the above dimensions the proposed Markov process method is more efficient than the universal generating function in aspects of evaluating the system availability.