Lysine is usually taken up in excess by the mammary gland (MG) relative to milk protein output, allowing for mammary synthesis of non-essential (NE) amino acids (AA) from Lys-N. It is unclear whether this NEAA synthesis from Lys is obligate or whether more efficient use of Lys can be made under limiting conditions. Six multi-catheterized dairy cows received a basal diet low in protein plus an abomasal infusion of AA (560 g/day) with or without Lys (50.3 g/day), in a crossover design with 7-day periods. On day 7, all cows received a 7.5-h jugular infusion of [2-15N]lysine. Six blood samples were collected from arterial, portal, hepatic and mammary vessels at 45 min intervals. In addition, cows were milked at 6 and 7 h with the milk casein plus arterial and mammary plasma collected at 7 h analyzed for AA enrichment. Milk protein concentration and casein yield tended (P < 0.10) to decrease with Lys deletion, while Lys secretion in milk protein was lowered (P < 0.05). The addition of Lys in the AA mixture increased the net portal absorption of Lys by the amount infused, suggesting limited oxidation of this extra supply by the gut. Net liver flux of Lys was unaltered by treatment and, therefore, net splanchnic release of Lys reflected closely the amounts absorbed. For both treatments, however, post-liver supply was greater than mammary uptake, which exceeded milk output. Nonetheless, while Lys deletion decreased mammary uptake by 10.1 mmol/h, Lys in milk protein secretion was reduced by only 3.9 mmol/h. On a net basis, there was no evidence of the additional uptake of any other measured AA during the Lys deletion. The mammary uptake to output ratio of Lys decreased from 1.37 to 1.12, but still showed an excess with Lys deletion. The total amount of 15N in milk protein did not change with treatment but the distribution into AA was altered. In conditions that simulated normal feeding (Lys infused), 83% of the 15N was present as Lys, with Glx, Asx, Ser and Ala harvesting, respectively, 6.8%, 2.4%, 2.1% and 1.0%. With Lys depletion, N-transfers from Lys to other AA within the MG were still present, but rates were considerably lower. This would suggest that part, at least, of Lys catabolism in the MG is either needed or cannot be prevented completely, even at low supply of Lys. Such catabolism will provide N to support the synthesis of NEAA.