Using postsynaptically tethered calcium sensor GCaMP, we investigated spontaneous synaptic transmission at individual active zones (AZs) at the Drosophila (both sexes) neuromuscular junction. Optical monitoring of GCaMP events coupled with focal electrical recordings of synaptic currents revealed "hot spots" of spontaneous transmission, which corresponded to transient states of elevated activity at selected AZs. The elevated spontaneous activity had two temporal components, one at a timescale of minutes and the other at a subsecond timescale. We developed a three-state model of AZ preparedness for spontaneous transmission and performed Monte Carlo simulations of the release process, which produced an accurate quantitative description of the variability and time course of spontaneous transmission at individual AZs. To investigate the mechanisms of elevated activity, we first focused on the protein complexin, which binds the SNARE protein complex and serves to clamp spontaneous fusion. Overexpression of Drosophila complexin largely abolished the high-activity states of AZs, while complexin deletion drastically promoted it. A mutation in the SNARE protein Syntaxin-1A had an effect similar to complexin deficiency, promoting the high-activity state. We next tested how presynaptic Ca 21 transients affect the states of elevated activity at individual AZs. We either blocked or promoted Ca 21 influx pharmacologically, and also promoted Ca 21 release from internal stores. These experiments coupled with computations revealed that Ca 21 transients can trigger bursts of spontaneous events from individual AZs or AZ clusters at a subsecond timescale. Together, our results demonstrated that spontaneous transmission is highly heterogeneous, with transient hot spots being regulated by the SNARE machinery and Ca 21 .