Phasic bursting by magnocellular neuroendocrine cells (MNCs) in the mammalian supraoptic and paraventricular nuclei (SON and PVN) consists of successive periods of action potentials and inactivity. It has previously been correlated with increased release of vasopressin from the neurohypophysis. In the present studies we investigated the neuronal mechanisms underlying this firing pattern. Using coronal slices of rat hypothalamus, we recorded intracellularly from neurons that are considered to be MNCs, based on several criteria. Eight of the 29 cells in this study displayed phasic burst patterns similar to those previously recorded extracellularly from MNCs in intact animals. Among the eight phasic cells, low levels of steady current injection could dramatically alter burst periodicity. Steady hyperpolarization revealed patterned synaptic input in only one case; in the remainder of the cells, nonsynaptic mechanisms appeared to account for periodic bursting. The phasic burst pattern usually appeared to be spike dependent, each burst arising from one or several depolarizing after-potentials (DAPs). Summed DAPs formed a plateau potential, which provided a depolarizing drive for further spiking. Spike frequency decreased late in the burst, and then the plateau potential terminated. During the quiescent period, burst excitability appeared to increase coincident with a small slow depolarization. Spikes and their summating DAPs could then initiate another burst. In several silent MNCs, a brief supra-threshold current pulse could initiate a prolonged afterdischarge, which had the properties of a phasic burst. Two MNCs that fired with a fast-continuous pattern were tested with brief hyperpolarizing current pulses; after each pulse, spike activity ceased and a plateau potential was revealed. Therefore, it appears that a maintained plateau potential (summed DAPs) can drive fast-continuous firing. In one case a periodic bombardment of excitatory postsynaptic potentials (EPSPs) generated a phasic firing pattern. The dependence of the burst characteristics on membrane potential, the apparent lack of patterned synaptic input in most cells, and the ability to evoke bursts with brief stimuli support the hypothesis that bursting in some MNCs involves an endogenous mechanism. Furthermore, phasic firing may be driven by tonic excitatory input. The data on phasic, silent, and fast-continuous cells suggest that many MNCs can generate DAPs and plateau potentials.