January 1984 Modified Contemporary Comparison sire evaluations for first, all, and later records were used to create sire-son pairs in five dairy breeds. Each evaluation for each bull included 10 or more daughters. Regressions of son on sire for evaluations for first, all, or later records exceeded the expectation of .5 and were most divergent for Jerseys and Brown Swiss. Regressions and coefficients of determination were generally highest when sire's evaluation based on all records predicted sons’ evaluations from first, all, or later records. Regression of son on sire for difference in evaluations (later minus first) was positive in all breeds, ranging from .08 to .10, and was significant for all breeds except Brown Swiss. Coefficients of determination were low (.01 to .02). Both regression of son on sire for difference and coefficient of determination increased with son's Repeatability, with a regression of .33 and a coefficient of determination of .09 for sons over 99% Repeatability, Regressions for 8,055 Holsteins with evaluations for sires and maternal grandsires on those ancestors were .46 to .54 for sires and .25 to .31 for maternal grandsires. Regressions of son's evaluation on pedigree index were .94 for first, 1.05 for all, and 1.12 for later lactations.