The most generally accepted production rate of Antarctic Bottom Water is 20 × 10 6 m 3/sec required by the STOMMEL and ARONS (1960) abyssal circulation model. According to this model, after sinking in the Weddell Sea, this water flows northward into the ocean basins and upwells through the thermocline. Antarctic Bottom Water is rich in dissolved silicon but the thermocline and surface waters are almost silicon-free. If the model is valid, it follows that a removal of silicon must take place from the upwelled waters before they return southward as required by mass conservation. In the Pacific, the amount of silicon removal necessary to satisfy the model is calculated for a transpacific (SCORPIO) section at 28°S. This calculated removal is 30 × 10 14g SiO 2/yr, which is many times the rate of silicon input to the world oceans. Two mechanisms for silicon removal north of 28°S are discussed, but reasons are given for rejecting both of them. It is concluded that production of Antarctic Bottom Water is negligible at the present time.