Although it is well accepted that vasopressin (ADH) increases the permeability to water of the toad bladder granular cell's luminal membrane, recent studies have suggested that regulation also takes place at an additional “postluminal” site within the epithelial granular cell. These studies are based upon the observation that a number of experimental maneuvers can alter tissue permeability to water, but do not change the number of particle aggregates observed on the protoplasmic face of the granular cell's luminal membrane with freeze‐fracture electron microscopy. These aggregates are believed by many investigators to mediate the transport of water across the luminal membrane.The dissociation between permeability and aggregate frequency described above has been variously interpreted as the consequence of changes in the permeability of the aggregates themselves, or of changes in the permeability of a “postluminal” barrier that is functionally in series with the luminal membrane.We attempted to distinguish between these 2 possibilities by studying paired toad bladders during 3 protocols that alter vasopressin‐stimulated water flow across the intact tissue without altering aggregate frequency. Estimates of the permeability of postluminal barriers were obtained by exposing the luminal surface to amphotericin B, an antibiotic that forms water‐permeant channels in the luminal membrane.Of the 3 protocols, only diminishing bladder filling volume decreased the water flow elicited by luminal amphotericin B, suggesting that only that protocol indeed decreased the permeability of some postluminal barrier. The other 2 protocols, increasing PCO2and repeatedly stimulating the bladder with vasopressin, did not alter amphotericin B‐elicited flow, suggesting that postluminal barriers were not altered by these 2 protocols. We thus conclude that a decrease in luminal membrane aggregate permeability or a decreased frequency of fusion of aggregate‐rich tubules to the luminal membrane mediates the decreased flow‐response in the latter 2 protocols, and not modulation of a postlumin