1. In the isolated perfused liver from 48h-starved rats, glycogen synthesis was followed by sequential sampling of the two major lobes. 2. The fastest observed rates of glycogen deposition (0.68-0.82mumol of glucose/min per g fresh liver) were obtained in the left lateral lobe, when glucose in the medium was 25-30mm and when gluconeogenic substrates were present (pyruvate, glycerol and serine: each initially 5mm). In this situation there was no net disappearance of glucose from the perfusion medium, although (14)C from [U-(14)C]glucose was incorporated into glycogen. There was no requirement for added hormones. 3. In the absence of gluconeogenic precursors, glycogen synthesis from glucose (30mm) was 0-0.4mumol/min per g. 4. When livers were perfused with gluconeogenic precursors alone, no glycogen was deposited. The total amount of glucose formed was similar to the amount converted into glycogen when 30mm-glucose was also present. 5. The time-course, maximal rates and glucose dependence of hepatic glycogen deposition in the perfused liver resembled those found in vivo in 48h-starved rats, during infusion of glucose. 6. In the perfused liver, added insulin or sodium oleate did not significantly affect glycogen synthesis in optimum conditions. In suboptimum conditions (i.e. glucose less than 25mm, or with gluconeogenic precursors absent) insulin caused a moderate acceleration of glycogen deposition. 7. These results suggest that on re-feeding after starvation in the rat, hepatic glycogen deposition could be initially the result of continued gluconeogenesis, even after the ingestion of glucose. This conclusion is discussed, particularly in connexion with the role of hepatic glucokinase, and the involvement of the liver in the glucose intolerance of starvation.