[en] A methodology is developed for identifying cost-effective failure prevention strategies for components or systems of nuclear power plants. Improvement of component reliability and addition of redundant components are identified as two basic strategies. Statistical decision theory is adopted as the tool for determining the optimum course of action among the strategies studied on the basis of minimum expected cost. A reliability analysis is developed as an important input to the decision model. Both repairable and nonrepairable systems are examined, and an example problem is developed for repairable systems. It is clearly shown that large expenditures are justified to improve the reliability of some major systems in nuclear power units and that these expenditures, even though they can exceed the original capital cost of the system, could save tens of millions of dollars per installation