The isolated intestinal mucosa of the flounder, Pseudopleuronectes americanus, when bathed in a 20 mM HCO3-Ringer's solution bubbled with 1% CO2 in O2, generated a serosa-negative PD and, when short-circuited, absorbed Cl at almost 3 times the rate of Na. Reducing HCO3 to 5 mM decreased the net Cl flux by more than 60%. The following results suggest that, despite the PD, Na and Cl transport processes are nonelectrically coupled: replacing all Na with choline abolished both the PD and net Cl flux; replacing all Cl with SO4 and mannitol abolished the PD and the net Na flux; and adding ouabain (to 0.5 mM) abolished the PD and the net Cl flux. Nearly all of the unidirectional serosa-to-mucosa Cl flux (JClsm) seemed to be paracellular since it varied with PD and Cl concentration in a manner consistent with simple diffusion. JClsm was only about one-fourth of JNasm, suggesting that the paracellular pathway is highly cation-selective. The data can be explained by the following model: (i) Na and Cl uptake across the brush border are coupled 1 : 1; Na is pumped into the lateral space and Cl follows passively, elevating the salt concentration there; (ii) the tight junction is permeable to Na but relatively impermeable to Cl; and (iii) resistance to Na diffusion is greater in the lateral space (considered in its entirety) than in the tight junction. If these assumptions are correct, the serosa-negative transmural PD is due mainly to a salt diffusion potential across the tight junction and, under short-circuit condition, most of the Na pumped into the lateral space diffuses back into the luminal solution, whereas most of the Cl enters the serosal solution. Morphological features of the epithelium support this interpretation: the cells are unusually long (60 micrometer); there is little distension of the apical 12 micrometer of the lateral space during active fluid absorption; and distension distal to this region is intermittently constricted by desmosomes.