Band 3-mediated Cl- exchange in human red blood cells and resealed ghosts was measured at 38 degrees C by the continuous flow tube method. When external Cl- concentration, C(o), is varied with constant internal Cl- concentration, C(i), the flux fits a simple Michaelis-Menten saturation curve (MM fit), with K1/2o = 3.8 +/- 0.4 mM. When the Cl- concentration is varied simultaneously at both sides of the membrane in resealed ghosts (C(i) = C(o) = C(i = o)), the flux rises toward a flat maximum between 200 and 450 mM Cl-, and then decreases at very high C(i = o). An MM fit to the data with C(i = o) < 500 mM gives K1/2s of 106 +/- 13 mM; fits including modifier site inhibition (MS fit) give an over threefold higher K1/2s. Despite this uncertainty, the intrinsic asymmetry of unloaded transport sites, A (defined as E(o)/E(i) with C(i) = C(o), where E(i) is the fraction of unloaded inward-facing sites and E(o) is the fraction of unloaded outward-facing sites), calculated from K1/2s and K1/2o, ranges only from 0.046 to 0.107. A new method, which uses the initial slope of a plot of Cl- flux versus C(i = o), gives A values of 0.023 to 0.038. Flufenamic acid (FA) inhibits Cl- exchange by binding to an external site different from the transport site. At 38 degrees C, FA binds 24-36 times more tightly to E(o) than to E(i). Estimates of A from FA inhibitory potency range from 0.01 to 0.05. All methods, including bicarbonate data from the preceding paper, indicate that at 38 degrees C, like 0 degree C, far more band 3 molecules are in the E(i) than in the E(o) form. The agreement of various methods supports the ping-pong model for anion exchange, and demonstrates that the intrinsic asymmetry is very slightly, if at all, affected by temperature.