At a variety of fast chemical synapses, spent synaptic vesicles are recycled via a large 'reserve' vesicle pool at high stimulus frequencies, and via fast 'local cycling' near release sites (e.g. 'kiss and run' transmitter release) at low stimulus frequencies. We have investigated recycling at the snake neuromuscular junction (NMJ), specifically seeking evidence for local cycling. Activity-dependent staining and destaining of the endocytic probe FM1-43 were directly compared to transmitter release over a range of stimulus frequencies. We found a fixed proportionality between staining/destaining and summed endplate potentials (EPPs) representing total transmitter release. There was no direct dependence of staining or destaining on stimulus frequency, as would be expected if local cycling (and consequent altered FM1-43 retention) were more prevalent at one frequency than another. In other experiments the drug vesamicol was used to abolish refilling of vesicles with transmitter, thereby blocking EPPs contributed by recycled vesicles. Control and vesamicol-treated NMJs had identical quantal content for the first 10 min of 1 Hz stimulation. Afterwards EPP amplitudes at vesamicol-treated NMJs declined at a rate consistent with use of a large pool containing approximately 130,000 vesicles. Finally, calibrated paired stimulations show that regenerated vesicles have poorer than random probability of re-release. Our findings are inconsistent with local cycling and suggest that the snake motor terminal utilizes exclusively a single large vesicle pool.