The present experiment aimed to evaluate the effect of HMBi on the production performance and metabolism in dairy cows. Thirty multiparous Holstein dairy cows under similar conditions were randomly assigned to three dietary treatments; i) Control, a basal diet; ii) T1, a basal diet plus HMBi (0 g prepartum and 18 g postpartum); and iii) T2, a basal diet plus HMBi (10 g prepartum and 18 g postpartum). Treatments were initiated 21 d before expected calving and continued through 91 d postpartum. HMBi was top-dressed onto the total mixed ration of each cow. Treatments did not affect dry matter intake, plasma urea nitrogen, peak milk yield, days to peak milk yield, nonesterified fatty acid, glutamate pyruvate transaminase, glutamic oxalaetic transaminase, milk fat content, milk protein content, milk lactose content, and milk solid non-fat content. The milk composition yields were increased by the HMBi-supplemented treatment. The T1 and T2 treatments increased the yields of 4% fat-corrected milk yield, milk fat, milk protein, and milk lactose compared with the control. Although there was no difference in the milk composition of the control and T2-treated cows, the T2-treated cows exhibited higher milk fat yield (increased by 74 g/d), lower milk urea nitrogen (reduced by 3.41%), and plasma ${eta}$-hydroxy butyrate than the control cows. The results indicate that HMBi supplementation to diet has beneficial effects, and that there is no difference between supplementation at prepartum and starting only at parturition.