In the acinar cells of the rat parotid gland the two membranes participating in exocytosis, i.e., the luminal plasmalemma and the secretory granule membrane, are clearly distinguishable in freeze-fracture because of their different densities in particles. In order to obtain point-specific information about the fusion-fission of these two membranes that occurs during the secretory cycle, glands were studied at various times (5 min to 6 h) after stimulation with isoproterenol. We observed that, in the course of the release of secretion products and shortly afterwards, the enlarged luminal plasmalemma exhibits a mosaic organization consisting of an alternation of membrane patches of high (original plasmalemma) and low (fused granule membrane) particle density. The transition between these two patterns is usually sharp. Later, concomitant with the reformation of acinar canaliculi, the low particle density membrane is found at the cell surface but only bounding vacuolar infoldings, and then it finally disappears.These results suggest that (a) fusion of these membranes does not result in a random intermixing of the molecular components of the participating membranes, which retain their structural identity; and (b) the enlarged luminal plasmalemma reverts to its original size by a progressive, specific removal of the regions of low particle density from the cell surface.In secretory cells the exportable proteins are known to be vectorially transported from the site of synthesis to the extracellular space through a pathway which includes a number of distinct membrane-bounded organelles (8,9,23,24): in sequence, the rough endoplasmic reticulum, the Golgi complex, and the secretion granules. The content of the latter is then discharged by exocytosis, i.e., by fusion of the granule limiting membrane with the plasmalemma followed by an opening at the point of fusion (36). Since the transport of the secretion products between adjacent cell compartments occurs in bulk, by means of membrane-bounded vesicles, a concomitant transfer of membranes must be postulated.The fate of the transferred membrane is still debated. Biochemical studies carried out on various secretory systems have clearly shown that the turnover rate of the molecular components of all the membranes involved is much slower than that of the secretory products (28,29,46,47); thus, it is most likely that these molecular components are reutilized in several secretory cycles (28,29). However, the mechanisms of this reutilization are still unclear, since it is not known whether membrane patches, after fusing with a membrane of a