1. A modification of the methods of Miller and of Schimassek for the perfusion of the isolated rat liver, suitable for the study of gluconeogenesis, is described. 2. The main modifications concern the operative technique (reducing the period of anoxia during the operation to 3min.) and the use of aged (non-glycolysing) red cells in the semi-synthetic perfusion medium. 3. The performance of the perfused liver was tested by measuring the rate of gluconeogenesis, of urea synthesis and the stability of adenine nucleotides. Higher rates of gluconeogenesis (1mumole/min./g.) from excess of lactate and of urea synthesis from excess of ammonia (4mumoles/min./g. in the presence of ornithine) were observed than are likely to occur in vivo where rates are limited by the rate of supply of precursor. The concentrations of the three adenine nucleotides in the liver tissue were maintained within 15% over a perfusion period of 135min. 4. Ca(2+), Na(+), K(+), Mg(2+) and phosphate were found to be required at physiological concentrations for optimum gluconeogenesis but bicarbonate and carbon dioxide could be largely replaced by phosphate buffer without affecting the rate of gluconeogenesis. 5. Maximal gluconeogenesis did not decrease maximal urea synthesis in the presence of ornithine and ammonia and vice versa. This indicates that the energy requirements were not limiting the rates of gluconeogenesis or of urea synthesis. 6. Addition of lactate, and especially ammonium salts, increased the uptake of oxygen more than expected on the basis of the ATP requirements of the gluconeogenesis and urea synthesis.