First lactations were from the North Carolina Institutional Breeding project since 1955 for cows and their dams where Modified Contemporary Comparisons were available for sires of both females. Each sire of cows was required to have a minimum of 20 progeny in the data. A total of 1771 cow-dam pairs by 33 sires in 7 herds and 217 herd-year-seasons passed all edits. Models including combinations of Predicted Difference milk for sire and maternal grandsire, dam's milk yield deviated from herd-year-season mean, and dam's age at cow's birth were used to predict cow's milk yield deviated from herd-year-season mean. Adjustment of yields of cow and dam for days open also was examined. Predicted Difference milk of her sire was the most important variable for predicting cow's yield; next was dam's yield. Predicted Difference of maternal grandsire and dam's age at cow's birth were not significant. Adjustment for days open reduced the percent of variation of cow's yield explained by sire's Predicted Difference. Sire's contributions to yield also were obtained by least squares and mixed model procedures in the various models with Predicted Difference milk of sire omitted. Maximum change in ranks among the 33 sires was four places, and maximum change of differences among sires was 129 kg when mate's yield and mate's sire's Predicted Difference were included in the model. Dam's yield was more useful than maternal grandsire's merit in predicting cow's yield. Differences among sires changed as much as 185 kg when cow's yield was adjusted for days open.