First available appraisals for 67644 Jersey cows were used to estimate relative magnitudes of additive and selected sources of nonadditive genetic variation, and evaluate corrective mating programs for 13 linear type traits scored linearly from 50 to 99 points. Estimates of additive genetic variances were 11 to 36 percent of within herd-classifier variance, with largest values for stature, strength, rump angle, rump width and udder depth. Dominance components ranged from 2 to 4 times as great as additive components for dairy character, rump width, and the udder traits; fore udder, rear udder height, rear udder width, and suspensory ligament, but were small for the other 7 traits. Maternal effects and the direct additive - additive maternal genetic covariance were unimportant. Sire predicted differences and four measures of dam's merit (dam's phenotypic score, this score relative to herd average, dam's cow index, and half the maternal grandsire predicted difference) were used to predict offspring scores, deviated from herd average. Sire regression coefficients were much larger than those for dam's measures, suggesting that sire evaluations should receive more weight than dam measures in a corrective mating context. Mean squares for effects on offspring scores of interactions between categorical levels of sire predicted differences and dam measures generally were nonsignificant, indicating that additively combining measures is appropriate for predicting offspring performance. Dam's cow index was found to be superior to dam's score for predicting mean performance of offspring groups (e.g., for large groups of offspring or over time) and is therefore preferred for corrective mating decisions. / Master of Science