Calcium release by photolysis has been applied to study mechanisms of release at large, multisite synapses (1,2). However, synapses made on the dendrites of central neurons, comprising a single or few release sites, may differ from these and may also show differences from one to another. Their small size is a technical impediment to studies of transmitter release. We describe here the application of laser photolysis to release Ca2+ in presynaptic terminals to study the parameters of vesicular release at single site connexions between cerebellar molecular layer interneurons (MLI). MLI have advantages for studies of this kind. Synapses between MLI show evidence of multivesicular release of GABA at single contacts (3). They are electrically compact, show a high likelihood of synaptic connection in recorded pairs, and have autoreceptors and autapses permitting single cell studies. Laser photolysis of DM-nitrophen 4 was used to release Ca2+ in the presynaptic terminal in paired whole cell recordings. The presynaptic interneuron was filled with Alexa 594 and Oregon Green Bapta 5N and the postsynaptic with Alexa 488. To find putative sites of connection and for mapping their morphologies, the preparation was illuminated alternately with 470/40 nm and 573/30 nm light from LED light sources, and imaged with an EMCCD camera through dual band filters. Photolysis of DM-Nitrophen-Ca was with a diode laser at 405 nm with controlled intensity, duration and frequency of exposure. The laser spot was set to 3 µm to illuminate single connexions and the preparation positioned with a piezo-controlled platform with resolution <0.25 µm. Ca2+ release was monitored by EMCCD imaging at 200-1000 Hz with 470/40 nm excitation and 535/40 nm emission. The method permits the timing and number of release events to be monitored postsynaptically with controlled pulses of presynaptic Ca2+ release. Data were analysed by measuring intervals between the laser pulse and the postsynaptic events in successive 1 s sweeps separated by 90 s. The distribution of intervals, fitted with maximum likelihood (5), showed two exponential components with time constants of 4 ms and 320 ms, and 60% of events were in the early component. This corresponds to a mean of 1.76 events /stimulus in the fast component. Corrections were applied for missed second and subsequent events due to postsynaptic receptor saturation, giving an estimate of the mean readily releasable pool of 2.0 /stimulus. The number of events varied from sweep to sweep at the same laser energy, up to a maximum observed of 4, and the probability of release is estimated as 0.7. The distribution of first latencies showed an initial silent period of 0.5 ms followed by a rapid rise to a maximum release rate in 1 ms, consistent with highly cooperative activation by photoreleased Ca2+.