Under conditions of chronic supply the liver removes most amino acids (AA) in excess of net anabolic needs. Little information is available, however, on how acute alterations in AA supply (as might occur with once-daily feeding regimens) are controlled by the liver. Are these also extracted completely in a ‘first-pass’ manner or are there limitations to hepatic uptake? Furthermore, is the rate of removal ‘saturable’ (by Michaelis–Menten kinetics) over the range of supply experienced under normal feeding conditions? These questions have been addressed in a study that involved acute (4.5 h) increases in AA supply. Four sheep were prepared with trans-hepatic vascular catheters and were offered a basal diet (equivalent to 1.6×energy maintenance) throughout. On four occasions, at 7 d intervals, they were infused with various amounts of an AA mixture into the mesenteric vein over a 4.5 h period. The mixture contained fourteen AA in the proportions present in rumen microbial protein. The amounts infused were calculated to provide an additional one, two, three and four times that absorbed from the basal diet. Continuous blood collections were removed over 2 h intervals before (basal diet only) and at 0.5–2.5 and 2.5–4.5 h of AA infusion. Transfers of AA, from the digestive tract and to the liver, were calculated for both plasma and total blood. The recovery of the infused AA across the portal-drained viscera (PDV) was quantitative (100%) only for histidine and proline, the remaining AA were recovered at 56–83 %. These losses correlated (P <0.001) with the arterial concentrations and were probably due to removal of AA from the systemic circulation by the tissues of the digestive tract. Despite the wide range of net PDV appearances (i.e. absorbed plus infused), the percentage of most AA removed by the liver remained constant, but the percentage varied with AA (from 34 for proline to 78 for tryptophan for blood transfers). Thus, even when supply was increased 5-fold over baseline there was no indication that the transport into the liver declined, indeed the absolute removals continued to increase. In contrast, the branched-chain AA (isoleucine, leucine and valine) did not exhibit constant percentage extractions. Their percentage extractions were always the lowest (16, 10 and 25 respectively) and tended to decline at the highest infusion rates, indicative of saturation in hepatic transport and/or metabolism. The arterial concentrations of all infused AA increased (P <0.001) with rate of infusion, again indicative that the liver did not extract all the net AA available across the PDV. Absolute amounts removed were similar between plasma and blood, indicating that most of the hepatic transfers occurred from plasma. The fractional rates of transfer from total inflow to the liver (i.e. with re-circulated AA included) were 3- to 4-fold lower than rates based on the amounts absorbed plus infused. The highest percentage extraction for total blood inflows was for serine (27), but most were between 6 and 16, except for the branched-chain AA, which were all <1. Use of percentage extractions based on total inflows are probably more appropriate for development of mathematical models of liver metabolism, and the current data suggest that constant values may be applied. The needs of the liver for specific mechanisms involving phenylalanine and histidine (plasma protein synthesis), glycine (detoxification of xenobiotics) and alanine (gluconeogenesis) probably also require to be included in such models.