Phasic activity in magnocellular neurosecretory vasopressin cells is characterized by alternating periods of activity (bursts) and silence. During phasic bursts, action potentials (spikes) are superimposed on plateau potentials that are generated by summation of depolarizing after-potentials (DAPs). Burst termination is believed to result from autocrine feedback inhibition of plateau potentials by the -opioid peptide, dynorphin, which is co-packaged in vasopressin neurosecretory vesicles and exocytosed from vasopressin cell dendrites during phasic bursts. Here we tested this hypothesis, using intracellular recording in vitro to show that -opioid receptor antagonist administration enhanced plateau potential amplitude to increase post-spike excitability during spontaneous phasic activity; the antagonist also increased post-burst DAP amplitude in vitro, indicating that endogenous dynorphin probably reduces plateau potential amplitude by inhibiting the DAP mechanism. However, the -opioid receptor antagonist did not affect the slow depolarization that follows burst termination, suggesting that recovery from endogenous -opioid inhibition does not contribute to the slow depolarization. We also show, by extracellular single-unit recording, that that there is a strong random element in the timing of burst initiation and termination in vivo. Administration of a -opioid receptor antagonist eliminated the random element of burst termination, but did not alter the timing of burst initiation. We conclude that dendritic dynorphin release terminates phasic bursts by reducing the amplitude of plateau potentials to reduce the probability of spike firing as bursts progress. By contrast, dendritic dynorphin release does not greatly influence the membrane potential between bursts and evidently does not influence the timing of burst initiation.