This paper describes component level reliability models that use infor
mation on degradation to predict component reliability, and which have
been used to evaluate different maintenance and testing policies. The
models are based on continuous time Markov processes, and are a gener
alization of reliability models currently used in Probabilistic Risk A
ssessment. An explanation of the models, the model parameters, and an
example of how these models can be used to evaluate maintenance polici
es are discussed.