AbstractThis survey paper considers possibilities ot usin&pattern recognition methods for prediction of product individual life time, rejection of potentially unreliable products and selection of those with different levels of reliability. Various pattern reoognition algorithms are described and their applicability to the problem of reliability forecasting modern oomponents is analysed. Also investigated are methods tor selection ot intormation-bearing sets of product parameters.