1. The modulation of the gamma-aminobutyric acidA (GABAA) receptor by reduced metabolites of progesterone and deoxycorticosterone has been compared with that produced by depressant barbiturates in: (a) voltage-clamp recordings from bovine enzymatically isolated chromaffin cells in cell culture, and (b) an assay of the specific binding of [3H]-muscimol to a preparation of porcine brain membranes. 2. The progesterone metabolites 5 alpha- and 5 beta-pregnan-3 alpha-ol-20-one (greater than or equal to 30 nM) reversibly and dose-dependently enhanced the amplitude of membrane currents elicited by locally applied GABA (100 microM), and over the concentration range 30 nM-100 microM stimulated the binding of [3H]-muscimol. In contrast, 5 alpha- and 5 beta-pregnan-3 beta-ol-20-one (30 nM-100 microM) had little effect in either assay, indicating a marked stereoselectivity of steroid action. 3. Scatchard analysis of the ligand binding data suggested an apparent increase in the number, rather than the affinity, of detectable [3H]-muscimol binding sites as the principle action of the active steroid isomers. 4. GABA-evoked currents were also potentiated by androsterone (1 microM) and the deoxycorticosterone metabolite 5 alpha-pregnane-3 alpha,21-diol-20-one (100 nM). 5. Secobarbitone (10-100 microM), pentobarbitone (10-300 microM) and phenobarbitone (100-500 microM) reversibly and dose-dependently potentiated the amplitude of GABA-evoked currents in the absence of any change in their reversal potential. 6. At relatively high concentrations (greater than or equal to 30 microM) secobarbitone and pentobarbitone directly elicited a membrane current. It is concluded that such currents result from GABAA receptor-channel activation since they share a common reversal potential with GABA-evoked responses (approximately 0 mV), are reversibly antagonized by bicuculline (3 microM), and potentiated by either diazepam (1 microM) or 5 beta-pregnan-3 alpha-ol-20-one (500 nM). 7. Secobarbitone (1 microM-1 mM) dose-dependently enhanced the binding of [3H]-muscimol. In common with the active steroids, an increase in the apparent number of binding sites was responsible for this effect. 8. A saturating concentration (1 mM) of secobarbitone in the ligand binding assay did not suppress the degree of enhancement of control binding produced by 5 beta-pregnan-3 alpha-ol-20-one (30 nM-100 microM). Similarly the steroid, at a concentration of 100 microM, did not influence the enhancement of [3H]-muscimol binding by secobarbitone (1 microM-1 mM). In all combinations of concentrations tested, the effects of secobarbitone and 5#-pregnan-3a-ol-20-one on [3H]-muscimol binding were additive. 9. In conjunction with previously published observations, the present data indicate close similarities in the GABA-mimetic and potentiating actions of barbiturates and steroids. However, the results obtained with combinations of steroids and barbiturates in the ligand binding assay appear inconsistent with the two classes of compound interacting with a common site to modulate the GABAA receptor activity.