Abstract
In this article we introduce the problem of computer software reliability and discuss a probabilistic model for describing the failure of software. We suggest a procedure for estimating the parameters of the model and propose a stopping rule for debugging the software. We apply our procedures to some published data on software failures.