We studied unidirectional [C-14]HCO3- efflux from human resealed red cell ghosts with 1 mm acetazolamide under self-exchange conditions at pH=pH((i=o)) 7.4-9.0 and 0-38 degrees C by means of the Millipore-Swinnex and continuous flow tube filtering techniques. (CO2)-C-14 loss from cells to efflux medium and further to the atmosphere was insignificant. [C-14]HCO3- efflux was determined at pH 7.8, 38 degrees C under symmetric variation of the HCO3- concentrations (C-(i=o), and asymmetric conditions: C-(i), varied, C-(o), constant, or C-(o), varied, C-(i), constant. MM-fit, J(D)(eff)=J(max)(eff). C .(C+K-1/2)-1, used to describe the concentration dependence of J(eff) when only C-(o) varied, yields at C(i)=50 mM: K-1/2(o)=3.8 mMJ, J(max)(eff,o)=20 nmol cm(-2) s(-1); at C-(i)=165 mM:K-1/2(o)=10 mM, J(max)(eff,o)=32 nmol cm(-2) s(-1). When ((i)) varied, noncompetitive self inhibition by HCO3- binding (inhibitor constant K-l) to an intracellular site was included (MS-fit). Under conditions of (a) symmetry: C-(i=o)=9-600 mM, K-1/2(s)=173 mM, K-l=172 mM, and J(max)(eff,s)=120 nmol cm(-2) s(-1), (b) asymmetry: C-(o)=50 mM, K-1/2(i)=116 mM, K-I=136 mM, and J(max)(eff,i)=92 nmol cm(-2) s(-1). All flux parameters accord with the ping-pong model for anion exchange. The data for C-(i) <200 mM, and J(max)(eff,s)=120 nmol cm(-2) s(-1), (b) asymmetry: C-(o)=50 mM, K-1/2(i)=116 mM, K-I=136 mM, and J(max)(eff,i)=92 nmol cm(-2) s(-2). All flux parameters accord with the ping-pong model for anion exchange. The data for C-(i) <200 mM also fit well to the MM equation, but K-1/2 and J(max)(eff) are different from the MS-fit and are inconsistent with the ping-pong model. Thus, self-inhibition (MS-fit) must be included even at low concentrations. As at 0 degrees C, the system is asymmetric: 8-10 times more unloaded transport sites face inward than outward when C-(i=o). J(eff,s) was not monoexponentially dependent on temperature at 0-38 degrees C, indicating that the transmembrane anion transport is controlled by several rate constants with different temperature dependencies. J(eff,s) was not significantly affected by increasing pH((i=o)), from 1.4 to 7.8, but it decreased by 50% when pH was raised to 9.0.