Abstract The wide range of transport rates for anions of differing chemical structure by the human erythrocyte anion transport protein (Band 3 protein) suggests that this protein is highly selective for anions that chemically resemble its natural substrate bicarbonate. To test this hypothesis, the influx of bisulfite (HSO3−1), a bicarbonate analog, was compared to influxes of chloride, sulfate, and bicarbonate, as measured by the technique of colloid osmotic lysis in isotonic ammonium salt solution. The lysis time induced in chloride solution (⪢ 10 min) was markedly accelerated to 0.6 min by the addition of small amounts (5 mM) of bicarbonate, an effect characteristic of colloid osmotic lysis induced by the anion transport pathway. Lysis in bicarbonate solution was extremely rapid (0.09 min), and was markedly inhibited by acetazolamide (2.9 min). Lysis in bisulfite solution occurred spontaneously (2.2 min) but was markedly accelerated to a time similar to that of chloride (0.56 min) by addition of 5 mM bicarbonate. In contrast, sulfate induced lysis was extremely slow (< 10% lysis at 40 min in the presence of bicarbonate). Preincubation of erythrocytes with SITS, an inhibitor of anion exchange, prevented lysis by chloride, but had no effect on lysis by bicarbonate, indicating that lysis by bicarbonate was predominantly through diffusion and not anion transport. SITS treatment of erythrocytes eliminated the catalytic effect of bicarbonate during lysis by bisulfite, indicating that anion transport of bisulfite and diffusion of the conjugate acid in the form of SO2 both contribute to the total membrane flux. When the contribution of diffusion is taken into account, the rate of bisulfite influx through the anion exchange pathway is at least 100-fold faster than that for sulfate.