Synapses are rightly viewed as devices primarily specializing in rapid, synchronous exocytotic transmitter release closely linked to presynaptic electrical excitation by Ca2+ influx through selective voltage-dependent channels, and we are probably just beginning to understand the complex machinery which ensures that this link is rapid and temporally selective. Thus, following a presynaptic impulse, multiple vesicles at various presynaptic locations undergo exocytotic fusion so nearly simultaneously that the changes in postsynaptic membrane conductance mediated by the transmitter content of each vesicle (the so-called quantum) sum to produce a much larger, synchronous postsynaptic voltage or current response in which the quantal elements cannot be individuated. Responses to asynchronous release of individual quanta, on the other hand, were, of course, immediately observed and studied once synaptic potentials could be directly recorded (Fatt & Katz, 1952). Until now, two phenomenologically distinct types of asynchronous release have been distinguished. The first, spontaneous (or constitutive) transmitter release, persists in the absence of presynaptic excitation, i.e. when action potentials are blocked, and produces small and discrete postsynaptic responses in a strictly random time series, so-called miniatures or minis. In the classical view, their randomness in time reflects the probabilistic nature of quantal release, while a synchronous impulse-evoked postsynaptic response is simply due to a very short-lasting but vigorous increase in the probability of their occurrence triggered by voltage-dependent Ca2+ influx. (Katz, 1966). The idea that minis represent chance occurrences of the same quantal elements that sum to form the much larger, synchronous postsynaptic response following an action potential has, however, recently been challenged (reviewed in Sullivan, 2009). The second, called either ‘slow release’, ‘late release’ or ‘asynchronous delayed component of evoked release’ is a transient enhancement of the frequency of asynchronous quantal events following the fast, synchronous postsynaptic response to a presynaptic action potential and building up during trains of impulses. This kind of asynchronous release is, of course, abolished along with synchronous release when presynaptic excitation is made impossible. It is classically interpreted as a result of the buildup of residual Ca2+ in the terminal, but more recently, a second, high-affinity Ca2+ sensor has been implicated in this phenomenon, which may also be linked to Ca2+-dependent processes of short-term facilitation (Sakaba & Neher, 2008). With the debate on the mechanistic relationship of these two forms of asynchronous release with one another and with synchronous release ongoing, the report by Popescu et al. (2010) published in a recent issue of The Journal of Physiology introduces yet a third type of asynchronous release: this novel form is neither dependent on action potential firing or presynaptic voltage-dependent Ca2+ entry nor stochastic in the sense of carrying no temporal signature deviating from randomness, instead occurring in clusters or bursts. Recording in whole-cell patch clamp mode from magnocellular neurons in acute in vitro slice preparations of rat hypothalamus, in about half of their recordings the authors observed bursts of grouped GABAergic inhibitory postsynaptic currents (IPSCs) occurring on average once every 13 min (at 28°C) and lasting several seconds with a frequency of quantal-like IPSCs 10- to 100-fold over the stochastic background mini activity. Intriguingly, like the latter, the bursts were resistant to block of both voltage-dependent sodium and calcium channels by tetrodotoxin (TTX) and Cd2+, respectively. TTX did reduce their frequency by about 4-fold, halved their incidence, and abolished the occurrence of correlated bursts in simultaneously recorded neighbouring neurones, thus clearly showing some influence of neuronal activity. However, albeit at lower frequency, the IPSCs did continue to burst healthily in the absence of action potential firing and voltage-dependent Ca2+ entry. One notes with some relief, then, that there is one thing that does seem to affect them less subtly: Popescu et al. report that removal of extracellular Ca2+ abolished IPSC bursting altogether. Together with their finding that intracellular Ca2+ stores seem uninvolved, Ca2+ influx into the presynaptic GABAergic terminal through a pathway other than voltage-gated Ca2+ channels is apparently required to trigger bursts. Now the great question is: through which pathway does Ca2+ enter the terminal? The authors consider presynaptic Ca2+-permeable ligand-gated non-selective cation channels, of which a number have been described (Schicker et al. 2008). Of course, we know already that, wherever the ligand comes from, it cannot be from synaptic terminals operated in the conventional manner requiring both action potentials and voltage-dependent Ca2+ channels. A factor that is released by neurones or glial cells in a manner less strictly linked to neuronal activity would possibly be required to fit the bill. The second big question raised by this work concerns a possible role of the bursts in the control of the pulsatile output of oxytocin from the magnocellular axon terminals in the neurohypophysis during birth and lactation. More generally, their finding might be the first inkling of a novel mode of synaptic transmission independent of (or at least very loosely dependent on) neuronal excitability and possibly even of a specific functional mission of slow, asynchronous transmitter release in the operation of a synaptic circuit.