1. To examine the dependence of gamma-aminobutyric acid (GABAA) receptor gating on the alpha-subunit isoform, we studied the kinetics of GABA-gated currents (IGABA) of receptors that differed in the alpha-subunit subtype, alpha 1 beta 2 gamma 2S and alpha 3 beta 2 gamma 2S. cDNAs encoding rat brain subunits were co-expressed heterologously in HEK-293 cells and the resultant receptors studied with the whole-cell patch clamp technique and rapidly applied GABA pulses (5-10 s). 2. IGABA of both receptors showed a loosely similar dependence on GABA concentration over a wide range (1-5000 microM). Generally, IGABA manifested activation reaching an early current peak, subsequent slower spontaneous desensitization, and deactivation of open channels at pulse termination. Lowering GABA concentrations reduced peak currents and slowed activation and desensitization kinetics. 3. The presence of alpha 3 altered the peak IGABA concentration-response relationship by shifting the fitted Hill equation to tenfold greater GABA concentrations (GABA concentration at half amplitude: alpha 1, 7 microM; and alpha 3, 75 microM) without affecting Hill coefficients (alpha 1, 1.6; alpha 3, 1.5). These findings indicate a reduction in the apparent activating site affinity and are consistent with previous reports. 4. To investigate differences in gating, we normalized for apparent activating site affinities by analysing the time course of macroscopic gating at equi-activating GABA concentrations. The presence of alpha 3 slowed activation fourfold (time to current peak (means +/- S.E.M.): alpha 1, 1.2 +/- 0.06 s (2 microM); alpha 3, 4.7 +/- 0.5 s (20 microM)), desensitization nearly twofold (reciprocal of time to 80% decay: alpha 1, 2.5 +/- 0.48 s-1 (100 microM); alpha 3, 1.5 +/- 0.15 s-1 (1000 microM)) and deactivation threefold (monoexponential decay time constant: alpha 1, 0.22 +/- 0.026 s (2 microM); alpha 3, 0.68 +/- 0.1 s (20 microM)). 5. To gain an insight into the gating mechanisms underlying macroscopic desensitization, we extended a previous gating model of GABAA receptor single-channel activity to include a desensitization pathway. Such a mechanism reproduced empirical alpha 1 beta 2 gamma 2S activation, desensitization and deactivation kinetics. 6. To identify molecular transitions underlying the gating differences between alpha 1 beta 2 gamma 2S and alpha 3 beta 2 gamma 2S receptors, we explored parameter alterations of the alpha 1 beta 2 gamma 2S gating model that provided an accounting of alpha 3 beta 2 gamma 2S empirical responses. Remarkably, alteration of rates and rate constants involved in ligand binding alone allowed reproduction of alpha 3 beta 2 gamma 2S activation, desensitization and deactivation. 7. These results indicate that substitution of the alpha 3 subunit variant in an alpha 1 beta 2 gamma 2S receptor alters transition rates involved in ligand binding that underlie changes in apparent activating site affinity and macroscopic current gating. Furthermore, they argue strongly that the structural determinants of these functional features reside on the alpha-subunit.