Because 2,5-anhydro-D-mannitol (2,5-AM) seems to stimulate feeding by acting on the liver and because the hepatic membrane potential has been suggested to play an important role in control of feeding ("potentiostatic" hypothesis), we investigated the effect of 2,5-AM on the membrane potential of liver cells with microelectrodes using a superfused liver slice technique. 2,5-AM (2.5 mM), which reduces intracellular ATP in rat liver, hyperpolarized the liver cell membrane in mouse and rat liver slices by 4-7 mV. This hyperpolarization was reversed by quinine (1 mM), an unspecific blocker of Ca2+-dependent K+ channels, and abolished by apamin (20 nM), a blocker of Ca2+-activated K+ channels with low conductance. Amiloride at 10(-3) M, but not at 10(-6) M, or a low-Na medium (26 mM) also eliminated the hyperpolarization. The K+ channel blockers cetiedil (50 microM), glibenclamide (30 microM), and Ba2+ (5 mM); flufenamic acid (100 microM), a blocker of nonselective cation channels; and ouabain (1 mM), an inhibitor of the Na+-K+-adenosinetriphosphatase, did not significantly influence the 2,5-AM-induced hyperpolarization. It is concluded that 2,5-AM hyperpolarizes the liver cell membrane by activating Ca2+-dependent K+ channels. This activation seems to be impaired when the Na+/H+ exchanger is inhibited by amiloride or a low-Na+ medium. The findings also imply that the hyperphagic effect of 2,5-AM observed in rats is not associated with a decrease in the hepatic membrane potential, as postulated by the potentiostatic hypothesis.