To determine the mechanism and pathway of entry of polar nonelectrolytes into bile, we studied first the permeation of [3H]H2O, [14C]urea, [14C]erythritol, [14C]mannitol, [3H]sucrose, [3H]inulin and [3H]dextran across an isolated, in situ perfused segment of the guinea pig's extrahepatic bile duct. All of these molecules, except [3H]dextran, permeated the bile duct. The diffusive permeability coefficients (cm per sec per 106) ranged from 248 for [3H]H2O to 1.2 for [3H]inulin. On the basis of these results, we formulated two models of solute biliary excretion. In Model I, permeation across both the canaliculus and ductule/duct was assumed to be by simple diffusion. In Model II, solute was assumed to enter the canaliculus by convection and diffusion, and the ductule/duct by diffusion. Reflection coefficients and/or permeability coefficients for the canalicular membrane were then determined by fitting the equations describing these processes to the bile-to-plasma ratios of [14C] erythritol, [14C]mannitol, [3H]sucrose and [3H]inulin observed at different rates of bile flow produced by partially clamping the bile duct cannula and subsequently infusing taurodehydrocholate. Only when convection was included as a mechanism of canalicular permeation (Model II) was excellent fitting observed. In such a case, the reflection coefficients and permeability coefficients for the carbohydrates in question were similar to those reported for other transporting epithelia. Cholestasis produced by taurolithocholate (10 to 40 μmoles per kg, i.v.) was associated with an irreversible increase in both the sieving coefficient and permeability coefficient for [3H]sucrose and [3H]inulin, even when the inhibition of bile flow was fully reversible. The permeability to [14C] erythritol and [14C]mannitol was either not affected or minimally increased. These findings suggest that, in the guinea pig: (i) solutes as large as [3H]inulin enter the biliary tree both at the canaliculus and bile ductule/duct; (ii) [14C]erythritol and [14C]mannitol gain access into the canalicular lumen primarily by convection, whereas [3H]sucrose and [3H]inulin permeate mainly by diffusion; (iii) distal permeation of these carbohydrates is small, and accounts for 1 to 11% of their total biliary entry; (iv) the canalicular membrane permeability to [3H]sucrose and [3H]inulin is not functionally important in bile secretion, and (v) a fraction of canalicular bile flows through the transjunctional shunt pathway.