In the conventional genetic evaluation, the records without sire identification were discarded. Four alternative methods were examined accuracy of animal model genetic evaluations by including records lacking sire identification in the simulated data. The alternatives assumed that genetic values of sires missing sire identity were equivalent to population(method 1) and contemporary group(method 2) averages. The other two alternatives accounted genetic values of cows missing sire identity in the model with producing abilities of random (method 3) and fixed(method 4) effects. Discarding records without sire identity resulted poor estimation of contemporary group effects. Method 1 and 3 were uniformly more accurate in estimating cow and sire breeding values. And method 3 showed the greatest increase in accuracy.