A method is presented for deriving an empirical function of component hazard rates from unit time and failure histories. The probability distribution of the mean hazard rate over a selected interval is derived, together with an excellent approximation which may be used for obtaining confidence statements or conducting simulation studies. The method has been reduced to practice, including computer programs for the analysis of data and graphical display of the results. The method is applied to an item of airborne electronic equipment and the results discussed. Nomenclature X = estimator of X E(X) = expected value of X F(X) = cumulative distribution function of X H(t) = sample hazard integral Hk = increment of H(t) N(t) = sample size at unit age t Nk = mean sample size over the A;th interval C(T) = measure of hazard rate variation h(t) = unit hazard integral hk = increment of h(t) rk = number of failures occurring at the fcth failure point r(t) = cumulative number of failures occurring in the sample t = unit age tk = unit age at /bth failure point \(t) = unit hazard rate at age t T = unit age measured from beginning of useful life