Eight multiparous lactating Holstein cows were randomly assigned in a 4 × 4 replicated Latin square design to receive four dietary treatments. The dietary treatments were monosodium glutamate by-product (MSGB) replacement for soybean meal in concentrate at four levels: MSGB replacement at 0, 20, 40 and 60%, respectively. Pangola hay was given on an ad libitum basis. It was found that total dry matter intake, concentrate intake, pangola hay intake and all apparent digestibilities were not different among treatments. Ammonia nitrogen concentration in the rumen at 4 h post-feeding was significantly different, in which the 0% treatment had the highest (P < 0.05) while the 20% treatment had the lowest. Milk fat percentage was the highest (P < 0.05) in the 0% treatment. MSGB replacement at 40% and 60% were shown to be the lowest (P < 0.05) feed cost for milk production, and profitability of milk production was the highest (P < 0.05) for the 60% treatment. Based on this experiment, it could be concluded that MSGB replacement for soybean meal at 20-60% in the feed for dairy cows presented no negative effects on their performances. In addition, it could decrease feed cost 2.9-17.3% and increase milk production profit up to 33.3% in the 60% treatment.