The Problem of estimating the failure rate （or MTBF） of an electronic product under development testing is considered from a Bayesian approach

m sets of life trials are performed under conditions which lead to an ordering of failure rates λ1>λ2>…> λm. The parameter of interest is Am. the final underlying failure rate of the product. The marginal posterior probability density function for Am is obtained when the conjugated type prior distribution without prior knowledge is assumed. An approximate solution is derived to meet the practical engineering needs

and the method is illustrated.