Multiparous Holstein cows (n = 122) were used in a randomized block design (blocked by parity, calving date, and risk factors) to determine the effect of SBE on 305-d milk yield and milk markers of inflammation and metabolism. Treatments were (1) control (CON), (2) 5-d administration of SBE (SBE5), and (3) 60-d administration of SBE (SBE60). Treatments were included in a treatment pellet that was identical to the control pellet except for the extract, both provided via an automated milking system beginning on d 1 of lactation. Milk samples were collected on d 1, 3, and once during d 5–12 of lactation, followed by weekly sampling until 120 d in milk. Milk samples collected in the first 2 wk were analyzed for biomarkers (haptoglobin, BHB, and glucose-6-phosphate [G6P]) and all samples were analyzed for components. Milk production, programmed pellet allocation, and actual provision of both pelleted feeds were recorded daily. Data were analyzed using the MIXED procedure in SAS. Treatment effects were evaluated by contrasts between CON and SBE5 and CON and SBE60 for both the treatment (TP; wk 1–9) and carryover (CP; wk 10–37) periods. Total pellet allocated was greater for SBE60 in both the TP (P < 0.01) and CP (P = 0.02), but was not different for SBE5 in either period. Treatments did not affect milk haptoglobin, BHB, or G6P. Whole-lactation milk yields were 11,245, 11,608, and 12,664 ± 465 kg for CON, SBE5, and SBE60, with differences between SBE60 and CON (P = 0.03). SBE5 did not alter milk components. SBE60 increased lactose and fat yields (P < 0.05) and tended to increase protein yield (P = 0.09) during TP; each increased during CP (P < 0.05). Somatic cell count decreased in SBE60 during TP (P = 0.02) but not CP. Mastitis incidence was lesser for SBE5 (P = 0.04) and SBE60 (P = 0.05) vs. CON. SBE60 tended to decrease the hazard of leaving the herd compared with CON and SBE5 (P = 0.07). Overall, feeding SBE for 60 d postpartum increased 305-d milk yield, possibly related to decreased milk somatic cells.