1. Isolated desheathed rat superior cervical ganglia were incubated in [3H]2,3,-gamma-aminobutyric acid ([3H]GABA) solution (1--10 microM for 2--3 hr) in the presence of 10 microM-amino-oxyacetic acid (AOAA). The subsequent efflux of tritium into a stream of superfused non-radioactive GABA-free Krebs solution at 25 degrees C was measured. 2. In the presence of 10 micrometer-AOAA the mean basal efflux rate coefficient (k0) for exit of tritium into the superfusion fluid was 0.7 x 10(-3) min-1. More than 98% of effluent tritium comprised unchanged [3H]GABA. The rate coefficient showed no correlation with the amount of [3H]GABA previously accumulated by the ganglion. 3. Elevation of [K+]o to greater than 50 mM increased the rate coefficient for [3H]GABA release by up to four times. Changes in efflux rate were not correlated with osmotic changes, and persisted after re-accumulation of effluent [3H]GABA by the inward carrier was inhibited. The effect of alkali metal cations diminished in the order Rb+ greater than K+ greater than Cs+Li+. Effects of K+ solutions were not reduced by omitting Ca2+ ions, with or without the addition of Mg2+. 4. Application of electrical pulses (0.1--1 msec duration, 1--10 Hz, 4 min trains) to the ganglion soma or to the preganglionic nerve trunk also raised k0. This effect declined with repeated stimulus trains, without an accompanying diminution in the response to K+. Responses to electrical stimulation were not reduced by amethocaine (300 microM), tetrodotoxin (3 microM) or raised [Mg2+i1 (0 mM-[Ca2+]/30 mM-[Mg3+]). Separate local superfusion of the pre- and post-ganglionic nerve trunks and of the ganglion soma showed that the response to electrical stimulation was localized to the vicinity of the stimulus and was not propagated along the nerve trunks or across the synapses. 5. Electrical recording from impaled 'inexcitable' cells (presumed to be neuroglial cells (Appendix)) indicated that the quantities of K+ ion accumulating during repetitive nerve stimulation are insufficient to stimulate the release of GABA from the glial cells. No physiological role for the release process in modulating neuronal excitability could be adduced.