Two experiments were conducted to test the hypothesis that responses of milk production to the abomasal infusion of casein or hydrolysates of casein can be influenced by the proportion of peptide-bound amino acids in the infusate. Eight lactating cows were used in two Latin square experiments with period lengths of 10 d. In Expt 1, the four cows were at a late stage of lactation and the four treatments were a basal diet of silage and a barley–soya supplement alone or with infusions into the abomasum of 180 g/d of sodium caseinate, an enzymic hydrolysate of casein or an acid hydrolysate of casein; all treatments supplied equivalent amounts of all the amino acids, this being achieved by addition of free amino acids as required. As infused, the three treatments contained 89 (caseinate), 40 (enzymic hydrolysate) and 15 (acid hydrolysate) % of their amino acids as peptides. The increase of milk production in response to infusion was small and there were no significant differences between the infusion treatments. In Expt 2, the cows received a basal diet of grass silage and a supplement containing feather meal as the main source of protein. Treatments were the basal diet alone and with three infusion treatments as in Expt 1, except that the amount infused was 230 g/d. The response to infusion was greater in this experiment, there were significant differences between infusions and the yield of milk protein with the infusion treatments was linearly related (P = 0·001) to the proportion of peptide-bound amino acids in the infusate. Moreover, there was a close inverse relation (P = 0·007) between the proportion of peptide-bound amino acids in the infusate and the concentration of total peptidebound amino acids in blood plasma.