Six replications in Experiment 1 and four replications in Experiment 2 of a 3 x 3 Latin square arrangement of treatments were used to compare soybean meal or soybean meal partially replaced with fish meal or a protein blend for response in intake, milk yield and composition, ruminal NH3 N, blood urea, and ruminal fermentation in lactating Holstein cows. The blend contained 30% corn gluten meal, 30% poultry by-products, 30% blood meal, and 10% feather meal. Periods were 28 d, and the first 7 d were used for adjustment. In addition to these protein sources, diets contained corn silage, alfalfa haylage, dried cracked corn, ground barley plus added fat, and a mineral and vitamin mixture. In Experiment 1, mean DMI was 24.4 kg, mean milk yield was 36.7 kg, mean fat percentage was 3.48%, and mean milk protein percentage was 3.06%; there were no significant differences. In Experiment 2, DMI was different for soybeans (22.6 kg) versus other sources (21.4 kg), but milk yield (32.1 kg) and fat (3.39%) and protein (2.87%) percentages did not differ among diets. In Experiment 1, ruminal NH3 N was greatest for cows consuming soybean diets (11.0 mg/dl) and lowest for cows consuming diets containing the protein blend (8.7 mg/dl). No differences in VFA were found. The lack of response to RUP can be explained by a rather high intake of a fermentable diet, which supplied sufficient absorbable AA according to the Cornell AA model.