Previous efforts to simulate mammary metabolism have focused on energy, mostly considering amino acids (AA) in aggregate. The main objective of this work was to build a model of mammary metabolism, based on data from arterio-venous difference studies, which considered AA in sufficient detail to predict yields of milk solids. The model contains 19 state variables and considers the removal of 37 metabolites from blood, including 22 AA. It is driven by blood flow and arterial concentrations, and outputs include milk protein, milk lactose, and three classes of milk fat (by chain length). The model was parameterized using a balance version of it and the mean observations from four arterio-venous difference experiments, with a limited number of assumptions, and evaluated against these experiments. In assembling the balance model, milk protein output was not predicted satisfactorily, as some essential AA were not present in quantities great enough to support the rates of milk protein synthesis observed experimentally. Tryptophan showed the greatest deficit, followed by tyrosine plus phenylalanine, methionine, and histidine. In addition, significant quantities of pyruvate were needed to synthesize serine, glycine, and alanine. The supply of alpha-ketoglutarate plus glutamate to synthesize proline and glutamine was provided in part by catabolism of arginine; the remainder was derived from catabolism of other AA and energetic substrates.