Arteriovenous glucose difference across the mammary gland of the lactating rat was used as an 'instantaneous' monitor of mammary glucose uptake. Plasma [glucose] and arteriovenous glucose difference varied according to whether Halothane, diethyl ether or sodium pentobarbitone anaesthesia was used. In pentobarbitone-treated rats a 60% glucose extraction in the fed state decreased to 5% after 18 h starvation, and recovered to 40% and 59% after 15 min and 60 min re-feeding respectively. The increase and decrease in plasma [fatty acids] and the depletion and restoration of hepatic glycogen mostly followed similar time courses. Re-feeding was accompanied by a brief surge of plasma [insulin]. Starved lactating rats showed a markedly greater capacity than age-matched virgin rats in the oral and intraperitoneal glucose tolerance tests. Mammary glucose uptake in the starved rat was significantly restored by oral or intraperitoneal glucose or by insulin, but not by acetoacetate or by heparin-induced elevation of plasma [fatty acids]. The role of insulin and of possible changes in mammary sensitivity to insulin in the return of mammary glucose uptake on re-feeding is discussed.