Our earlier report that perfusion of Al and Ga salts through rat brain ventricles abolished cerebrospinal fluid (CSF) production is incorrect. We now realize there are three conditions which, when simultaneously present, will give a false reading of the events taking place. These are: metal salts, low pH with subsequent rise in pH during perfusion and use of blue dextran as the marker. When we perfused 10 mM AlCl3 or GaCl3 in vivo, all three conditions were provided, and led to the misinterpretation that all CSF production had been eliminated. In doing experiments that prevented the subsequent rise in pH during perfusion, or in using [14C]dextran as the marker in vivo, we found that Al and Ga (acting as Lewis acids at pH 4.7 and below), and also hydrochloric, phosphoric and acetic acids of the same pH, reduce CSF production by approximately 33%. We attribute this to the lowering of pH eliminating the uncatalyzed reaction OH- + CO2----HCO3- in secretory cells of choroid plexus. Acetazolamide blocks the HCO3- formation catalyzed by carbonic anhydrase, and reduces CSF production 42%. These mechanisms are additive in that, when acetazolamide and either Al or acetic acid are given together, there is a total reduction of 64%. Therefore, at least 64% of CSF production is dependent on formation of HCO3-. The mechanisms accounting for the balance of CSF production are not yet defined clearly. A significant portion of this production is expected to relate to transport of Cl-.