Abstract
The computation of confidence interval is one of the system reliability estimations. The estimated means and the variances of component reliabilities are used to get the mean and variance of the system reliability and compute the confidence interval for the system reliability. This paper presents two generalized methods to estimate the system reliability and its confidence bounds when there are no failures of some component tests and failures for the other component tests. These methods can be applied to many systemdesign configurations and component time-to-failure distributions. The methods do not require any parametric assumptions for component reliability or time-to-failure. These approaches apply to any system design which can be decomposed into series and/or parallel connections between the components.