Synaptic transmission is characterized by a remarkable trial-to-trial variability in the postsynaptic response, influencing the way in which information is processed in neuronal networks. This variability may originate from the probabilistic nature of quantal transmitter release, from the stochastic behavior of the receptors, or from the fluctuation of the transmitter concentration in the cleft. We combined nonstationary noise analysis and modeling techniques to estimate the contribution of transmitter fluctuation to miniature inhibitory postsynaptic current (mIPSC) variability. A substantial variability (approximately 30%) in mIPSC decay was found in all cell types studied (neocortical layer2/3 pyramidal cells, granule cells of the olfactory bulb, and interneurons of the cerebellar molecular layer). This large variability was not solely the consequence of the expression of multiple types of GABA(A) receptors, as a similar mIPSC decay variability was observed in cerebellar interneurons that express only a single type (alpha(1)beta(2)gamma(2)) of GABA(A) receptor. At large synapses on these cells, all variance in mIPSC decay could be accounted for by the stochastic behavior of approximately 36 pS channels, consistent with the conductance of alpha(1)beta(2)gamma(2) GABA(A) receptors at physiological temperatures. In contrast, at small synapses, a significant amount of variability in the synaptic cleft GABA transient had to be present to account for the additional variance in IPSC decay over that produced by stochastic channel openings. Thus, our results suggest a synapse-specific contribution of the variation of the spatiotemporal profile of GABA to the decay of IPSCs.