Accurate evaluation of breading value of cow on the basie of its own early performance and/or performance of its relatives, is a pre-requisite to effective genetic improvement. The fact that the observed genetic change differs widely from expectod genetic change suggeete reappraisal of procedures of evaluation of breeding values of cows. The theoretical formulae, given by Young (1961), for estimation of the weights to be given to various amounts of information on the cow ,it's paternal h alf-sisters and dam are based on various assumptions. The heritability is assumed to be same for firs t and subsequent records. Also the repeatability is assumed to be same for the various repeated records. Not a ll these assumptions hold true in actual situations. It was, therefore, considered useful to evaluate the weightages to be given to the cows' own records in a ll possible combinations with available records of its dam and paternal h alf-sisters by fittin g multiple linear regressions of breeding value of cou, based on lifetim e production, on the above various combinations*