1. Whole cell voltage-clamp recording with recording pipette solutions of differing ionic composition was used to determine the ionic basis of the depolarizing gamma-aminobutyric acid (GABA) response. In the presence of 4-aminopyridine and excitatory amino acid receptor blockers, giant GABA-mediated postsynaptic currents (GPSCs) were recorded from CA3 pyramidal neurons in hippocampal slices from adult guinea pigs. With the GABAB component blocked, the GPSC was composed of an initial outward current (GABAA component) that peaked at 115 ms followed by a late inward current (GABAD component) that peaked at 400-600 ms. 2. Reduction of the intracellular concentration of potassium ([K+]i)resulted in no significant change in the reversal potential of the GABAD component of the GPSC, indicating that it is not a nonspecific cation current. 3. The HCO3- permeability of the channel mediating the GABAD response was assessed by using recording pipette solutions containing three different concentrations of bicarbonate ([HCO3-], 19, 49, and 102 mM). The reversal potential of the GABAD response shifted in the depolarizing direction as the HCO3- equilibrium potential was shifted in the depolarizing direction, indicating that the channel mediating the GABAD response is permeable to HCO3-. The reversal potential of the GABAD response was more sensitive to changes in recording pipette [HCO3-] than the reversal potential of the GABAA response, indicating that the GABAD response is carried by HCO3- to a greater extent than the GABAA response. 4. The outward current-inward current sequence of the biphasic GPSC was reversed to an inward current-outward current sequence by using a high [Cl-]/low [HCO3-] recording pipette solution (40 mM Cl-/6 mM HCO3-), indicating that the GABAA component is more sensitive to changes in [Cl-]i, and the GABAD component is more sensitive to changes in [HCO3-]i. 5. These data indicate that the GABAD component of the GPSC is predominantly carried by HCO3-. While this result supports the recently propsed chloride accumulation model, the model in its present form cannot explain the inward current-outward current polarity sequence of the GPSC recorded with the high [Cl-]/low [HCO3-] intracellular solution. The data obtained using that solution reveal the need for a more expansive chloride accumulation/ depletion model or for a model utilizing two distinct ionotropic GABA channels with different anion permeability ratios to account for the biphasic nature of the GPSC.