Abstract: This article develops a new method, named M-Bayesian credible limit method, to estimate reliability parameter. In the article, the M-Bayesian credible limit method of failure rate is derived for zero-failure data from products with exponential distribution. Relations between M-Bayesian credible limit and other classical confidence limits are discussed. Finally, the new method isapplied to a real zero-failure data set, and as can be seen, it is both efficient and easy to operate.

Key words: M-Bayesian credible limit, Classical confidence limit, Exponential distribution, Failure rate, Zero-failure data