1. Radioligand binding and patch-clamp techniques were used to study the actions of gamma-aminobutyric acid (GABA) and the general anaesthetics propofol (2,6-diisopropylphenol), pentobarbitone and 5 alpha-pregnan-3 alpha-ol-20-one on rat alpha 1 and beta 3 GABAA receptor subunits, expressed either alone or in combination. 2. Membranes from HEK293 cells after transfection with alpha 1 cDNA did not bind significant levels of [35S]-tert-butyl bicyclophosphorothionate ([35S]-TBPS) (< 0.03 pmol mg-1 protein). GABA (100 microM) applied to whole-cells transfected with alpha 1 cDNA and clamped at -60 mV, also failed to activate discernible currents. 3. The membranes of cells expressing beta 3 cDNAs bound [35S]-TBPS (approximately 1 pmol mg-1 protein). However, the binding was not influenced by GABA (10 nM-100 microM). Neither GABA (100 microM) nor picrotoxin (10 microM) affected currents recorded from cells expressing beta 3 cDNA, suggesting that beta 3 subunits do not form functional GABAA receptors or spontaneously active ion channels. 4. GABA (10 nM-100 microM) modulated [35S]-TBPS binding to the membranes of cells transfected with both alpha 1 and beta 3 cDNAs. GABA (0.1 microM-1 mM) also dose-dependently activated inward currents with an EC50 of 9 microM recorded from cells transfected with alpha 1 and beta 3 cDNAs, clamped at -60 mV. 5. Propofol (10 nM-100 microM), pentobarbitone (10 nM-100 microM) and 5 alpha-pregnan-3 alpha-ol-20-one (1 nM-30 microM) modulated [35S]-TBPS binding to the membranes of cells expressing either alpha 1 beta 3 or beta 3 receptors. Propofol (100 microM), pentobarbitone (1 mM) and 5 alpha-pregnan-3 alpha-ol-20-one (10 microM) also activated currents recorded from cells expressing alpha 1 beta 3 receptors. 6. Propofol (1 microM-1 mM) and pentobarbitone (1 mM) both activated currents recorded from cells expressing beta 3 homomers. In contrast, application of 5 alpha-pregnan-3 alpha-ol-20-one (10 microM) failed to activate detectable currents. 7. Propofol (100 microM)-activated currents recorded from cells expressing either alpha 1 beta 3 or beta 3 receptors reversed at the Cl- equilibrium potential and were inhibited to 34 +/- 13% and 39 +/- 10% of control, respectively, by picrotoxin (10 microM). 5 alpha-Pregnan-3 alpha-ol-20-one (100 nM) enhanced propofol (100 microM)-evoked currents mediated by alpha 1 beta 3 receptors to 1101 +/- 299% of control. In contrast, even at high concentration 5 alpha-pregnan-3 alpha-ol-20-one (10 microM) caused only a modest facilitation (to 128 +/- 12% of control) of propofol (100 microM)-evoked currents mediated by beta 3 homomers. 8. Propofol (3-100 microM) activated alpha 1 beta 3 and beta 3 receptors in a concentration-dependent manner. For both receptor combinations, higher concentrations of propofol (300 microM and 1 mM) caused a decline in current amplitude. This inhibition of receptor function reversed rapidly during washout resulting in a "surge' current on cessation of propofol (300 microM and 1 mM) application. Surge currents were also evident following pentobarbitone (1 mM) application to cells expressing either receptor combination. By contrast, this phenomenon was not apparent following applications of 5 alpha-pregnan-3 alpha-ol-20-one (10 microM) to cells expressing alpha 1 beta 3 receptors. 9. These observations demonstrate that rat beta 3 subunits form homomeric receptors that are not spontaneously active, are insensitive to GABA and can be activated by some general anaesthetics. Taken together, these data also suggest similar sites on GABAA receptors for propofol and barbiturates, and a separate site for the anaesthetic steroids.