The aim of this study was to understand milk yield variations in response to a balanced profile of 10 essential amino acids (EAA) at 2 levels of metabolizable protein (MP) supplies. Diets and infusions provided 14.5 vs 16.8 g/MJ for MP∙NEL-1 in INRA 2007 system and 128 vs 145 g/kg DM of CP, respectively in Low vs high MP. Four lactating dairy cows received 2 dietary treatments (low vs high MP) and 2 different duodenal infusions of 10 EAA (AA- and AA+), according to a 2×2 factorial design. The AA- and AA+ treatments provided (in % of MP) 1.7 vs 2.4% of Met, 5.8 vs 7.0% of Lys, 7.9 vs 9.0% of Leu and 1.9 vs 3.0% of His, respectively. Six blood samples were collected from carotid artery and mammary vein to calculate the uptakes of all the energetic nutrients using the Fick principle on Phe + Tyr for mammary plasma flow. Results were analysed using the MIXED procedure of SAS (2009) taking the effect of cows as random. Milk yield increased in AA+ vs AA- treatments (9.1 vs 8.6 kg/12 h∙per half udder, P=0.04), with no significant difference in lactose yield between the 2 treatments (442 vs 422 g/12 h∙per half udder). However, the water content in milk was higher in the AA+ vs AA- treatments (7.90 vs 7.45 kg/12 h∙per half udder, P=0.05) and explained by a parallel increase in minerals yield (69 vs 64 g/12 h∙per half udder, P=0.05). Mammary net uptake of glucose (GLC) tended to increase in AA+ vs AA- treatments (1,729 vs 1,562 mmol/h of C, P=0.06) as the difference between GLC minus lactose (504 vs. 391 mmol/h of C, P<0.1). This indicates that there was a shift in GLC partition towards ATP furnishing to support the increased milk protein yield (MPY; from 262 to 290 g/12 h∙per half udder; P=0.01). In conclusion, milk volume could vary differently than lactose yield and MPY contrary to the meta-analysis of Daniel and Sauvant and the predictions of INRA (2018) model as it depends on minerals in response to EAA supplies.