ABSTRACT In the field of nuclear power plants there is a great interest in the study of the characteristics of reliability, maintainability and availability of their equipment and its influence on the security and the economy of the plant. Therefore, the decision making on improvement and, even optimization, of testing and maintenance activities at nuclear power plants can take be better solved from the simultaneous consideration of criteria RAMS+C. Therefore, the process to find optimal tests and maintenance strategies should balance the goals achieved, based on RAMS+C criteria. The challenge of this thesis is, firstly, the need to develop new models to explicitly represent the impact of maintenance and test on RAMS+C criteria. Secondly, development of new optimization methods capable of properly handling both, the complexity of the new models developed, as a large number of decision variables may be involved in the optimization process based on multiple RAMS+C criteria, and also the presence of uncertainties that affect the decision-making process, associated to parameters, models or decision variable. Genetic Algorithms have been chosen to find a solution of the multicriteria optimization problem, as they have already shown a high efficiency in solving complex problems. The objective of this thesis focuses on the development of new models and methods necessary to undertake testing and maintenance optimization process based on RAMS+C criteria, and their application to nuclear plants safety related systems. There exists several approaches that model the behavior of the systems with time independence, i.e. through the use of average values for RAMS+C attributes have been proposed. Next stage, proposes new RAMS+C models to consider time in the behavior of the systems. . This requires, the customizations of genetic algorithms previously used in the optimizations to handle new criteria and decision variables, and also the integration of improvements in the algorithms Finally, new models and methods that consider the effect of uncertainties associated with the parameters, models and, and even the decision variables in the optimization process are proposed. All the new proposals are analyzed in different cases of application that demonstrate the feasibility and applicability of the new models and methods.