The objective of this study was to identify useful secondary traits to improve the prediction of the genetic ability of milk production in dairy cattle. Thirty-seven Holstein calves were injected with glucose at the rate of 0.15 g/kg body weight after 24 h fast. A total of 566 milk records in 238 Holstein cows were collected. Plasma glucose concentration reached a peak 10 min after the injection. The concentrations of plasma non-esterified fatty acids (NEFA) and total ketone decreased 20 and 30 min after the injection, respectively. The collected blood samples were classified into 3 time groups: T1, 10 to 30 min; T2, 40 to 60 min; and T3, 70 to 90 min after the glucose injection. Heritability estimates of the plasma glucose, NEFA and total ketone concentrations ranged from 0.60 to 0.97 among the 3 time groups (T1, T2, and T3). Genetic correlation between the plasma glucose concentration and milk yield ranged from 0.50 to 0.76 in T1, T2 and T3. Genetic correlation between plasma glucose concentration and fat yield in T1 and T2 was -0.43 and -0.35, respectively. Genetic correlations between plasma NEFA or total ketone concentrations and milk production ranged from -0.22 to -0.97. The use of the concentrations of plasma metabolites increased the reliability of breeding values for animals without milk records by 7.7 to 122.9%, suggesting that the determination of the concentrations of plasma metabolites may improve the accuracy of prediction of the genetic ability of milk production traits. The combination of fast and glucose injection provides a suitable technique for generating plasma metabolites to predict the genetic merit of milk production in dairy cattle.