The study was designed to develop an improved technique for perfusing the isolated caudal lobe of sheep liver. Twenty caudal lobes were perfused for 3-4 h, in a non-recirculating mode, with Krebs-Henseleit bicarbonate buffer. The perfusion system was designed to give a constant flow. The hepatic viability and functional normality of the perfused lobe were assessed by measuring the perfusion flow rate, pH, K+ efflux, O2 uptake, substrate uptake, gluconeogenesis from propionate and amino acids, and ureagenesis from ammonia and amino acids. Liver tissue was sampled for histological examination, as well as for the determination of liver glycogen and wet : dry weight ratio. The perfusion flow rate and pH were both stable throughout the perfusion. The potassium concentration in the effluent perfusate did not increase during the perfusion, suggesting that there was no loss of viability or hypoxia. The perfused lobe extracted more than 50 % of the O2 supply. The rate of oxygen consumption was comparable to the rate reported in vivo. The initial glycogen content was reduced by about 40 % after 4 h perfusion. The wet : dry weight ratio was 3.6, consistent with the absence of tissue oedema. Urea production was stimulated when NH4Cl (0.3 mM) was added to the medium but there was no significant increase in urea release when alanine (0.15 mM), glutamine (0.2 mM) or lysine (0.2 mM) was added. Urea production, however, increased by about 171 % when a physiological mixture of amino acids was added. Propionate (0.5 mM), alanine and glutamine stimulated glucose production but not lysine or the complete amino acid mixture. Glutamine release was lower than that reported in the rat liver. Changing the direction of flow also revealed an apparent difference between livers from sheep and rats in their metabolism of ammonia. The improved technique offers a simple practical and inexpensive approach to many problems in ruminant physiology and nutritional biochemistry. Experimental Physiology (2000) 85.5, 469-478.