1. We have used the whole-cell configuration of the patch-clamp technique to investigate the effects of neuroactive steroids on GABAA receptor-mediated synaptic transmission between rat hypothalamic neurones and pituitary intermediate lobe (IL) cells grown in coculture. In order to discriminate between possible pre- and postsynaptic sites of action, the effects of neurosteroids on GABAA receptor-mediated synaptic currents (IPSCs) were compared with those of GABAA currents (IGABA) triggered by local application of 50 or 500 microM GABA, which yielded approximately half-maximal and maximal responses, respectively. 2. In primary cultures of rat pituitary IL cells, allopregnanolone (5 alpha-pregnan-3 alpha-ol-20-one) reversibly potentiated IGABA in a dose-dependent manner with a threshold between 0.1 and 1 nM. At a concentration of 10 nM, allopregnanolone increased the response evoked by 50 microM GABA by +21.4 +/- 5.1% (n = 8), but had no effect on IGABA induced by 500 microM GABA. The beta-isomer of allopregnanolone, epipregnanolone (5 beta-pregnan-3 beta-ol-20-one, 10 nM), had no effect on IGABA at any concentration of GABA tested. 3. At concentrations lower than 10 microM, pregnenolone sulphate (5-pregnen-3 alpha-ol-20-one sulphate) did not significantly inhibit IGABA. However, at 10 microM, a systematic reduction of IGABA evoked by 50 and 500 microM GABA was observed, with mean values of -80 and -60%, respectively. This blocking effect was reversible and accompanied by a marked acceleration of decay of GABAA currents during the application of GABA. 4. In isolated pairs of synaptically connected hypothalamic neurones and IL cells, allopregnanolone (10 nM) augmented the mean amplitude of spontaneous IPSCs (sIPSCs) and electrically evoked IPSCs (eeIPSCs) by about 40% and increased the mean frequency of sIPSCs. Allopregnanolone (10 nM) also markedly increased the frequency of miniature IPSCs (mIPSCs) recorded in the presence of TTX (0.5 microM), but without modifying their mean amplitude. Epipregnanolone had no effect on the amplitude or frequency of sIPSCs. Neither epipregnanolone nor allopregnanolone modified the time to peak and decay time constants of GABAergic IPSCs. 5. Pentobarbitone (50 microM), a positive allosteric modulator of GABAA receptors, did not affect the amplitude of sIPSCs or eeIPSCs, but significantly increased the decay time constants of both types of IPSCs. Pentobarbitone had no effect on the frequency of sIPSCs. 6. Pregnenolone sulphate (10 microM) completely and reversibly blocked sIPSCs and eeIPSCs. Progressive block of IPSCs was correlated with a gradual decrease of the mean decay time constant. 7. Our results suggest that, under physiological conditions, allopregnanolone might be a potent modulator of GABAergic synaptic transmission, acting at both pre- and postsynaptic sites. The involvement of pregnenolone sulphate as a modulator of GABAergic IPSCs under physiological conditions is, however, more questionable. The mechanisms of action of both types of neurosteroids are discussed.