Neurons maintain a limited pool of synaptic vesicles which are docked at active zones and are awaiting exocytosis^,,,. By contrast, endocrine cells releasing large, dense-core secretory granules have no active zones, and there is disagreement about the size and even the existence of the docked pool. It is not known how, and how rapidly, secretory vesicles are replaced at exocytic sites in either neurons or endocrine cells. By using electron microscopy, we have now been able to identify a pool of docked granules in chromaffin cells that is selectively depleted when cells secrete. With evanescent-wave fluorescence microscopy, we observed single granules undergoing exocytosis and leaving behind patches of bare plasmalemma. Fresh granules travelled to the plasmalemma at a top speed of 114 nm s⁻¹, taking an average of 6 min to arrive. On arrival, their motility diminished 4-fold, probably as a result of docking. Some granules detached and returned to the cytosol. We conclude that a large pool of docked granules turns over slowly, that granules move actively to their docking sites, that docking is reversible, and that the ‘rapidly releasable pool’ measured electrophysiologically represents a small subset of docked granules.