Broecker <1982> has proposed that during the retreat of the Wisconsin ice sheets, atmospheric CO2 increased because of the extraction of phosphorus from the ocean as sea level rose. Using a time-dependent box-model, we examine the consequences of the phosphate extraction hypothesis over the last 120,000 years, assuming that Δ18O change in core V28-238 is analogous to sea level variation. The model takes into account the total CO2 and alkalinity balance in the deep sea and in the ¿pper ¿reservoir consisting of the surface ocean and atmosphere, which are assumed to be in equilibrium. Dissolution of deep-sea calcium carbonate sediment is assumed to respond to the supply of particulate carbonate from the ¿pper ¿box and to the dissolved carbonate-ion concentration of the deep sea. Assuming 1015 mole of phosphorus and 1017 mole of carbon were extracted during deglaciation, the predicted increase in pCO2 is 54 ppm. Variation of pCO2 follows the ice-volume forcing function with a 1000 yr lag, which is the input residence time of water in the deep sea. The accumulation (supply minus dissolution) of CaCO3 and the percent preservation (accumulation over supply) follow the derivative of the ice-volume function. Both are similar to the solution index and percent fragments of core V28-238. In addition, unusually good preservation is predicted during deglaciation due to carbon extraction. If phosphorus but not carbon is assumed to be extracted, the accumulation of CaCO3 varies little, remaining near a value equivalent to the river input.