The classical concept of the vesicular hypothesis for acetylcholine (ACh) release, one quantum resulting from exocytosis of one vesicle, is becoming more complicated than initially thought. 1) synaptic vesicles do contain ACh, but the cytoplasmic pool of ACh is the first to be used and renewed on stimulation. 2) The vesicles store not only ACh, but also ATP and Ca(2+) and they are critically involved in determining the local Ca(2+) microdomains which trigger and control release. 3) The number of exocytosis pits does increase in the membrane upon nerve stimulation, but in most cases exocytosis happens after the precise time of release, while it is a change affecting intramembrane particles which reflects more faithfully the release kinetics. 4) The SNARE proteins, which dock vesicles close to Ca(2+) channels, are essential for the excitation-release coupling, but quantal release persists when the SNAREs are inactivated or absent. 5) The quantum size is identical at the neuromuscular and nerve-electroplaque junctions, but the volume of a synaptic vesicle is eight times larger in electric organ; at this synapse there is enough ACh in a single vesicle to generate 15-25 large quanta, or 150-200 subquanta. These contradictions may be only apparent and can be resolved if one takes into account that an integral plasmalemmal protein can support the formation of ACh quanta. Such a protein has been isolated, characterised and called mediatophore. Mediatophore has been localised at the active zones of presynaptic nerve terminals. It is able to release ACh with the expected Ca(2+)-dependency and quantal character, as demonstrated using mediatophore-transfected cells and other reconstituted systems. Mediatophore is believed to work like a pore protein, the regulation of which is in turn likely to depend on the SNARE-vesicle docking apparatus.