Calculations of the 32Si/SiO2 ratio obtained in ocean models are compared with measurements of this ratio made as part of the GEOSECS program (Somayajulu et al., 1987, 1991). A major difference is found; while the models predict threefold to fivefold higher ratios in the deep Atlantic Ocean than in the deep Pacific and Indian oceans, no such difference is seen in the measurements. This points to a flaw in the measurements. Our modeling results suggest that the most interesting application for this 120-year half-life cosmogenic isotope is as a monitor of upwelling. However, considering the expense and difficulty associated with 32Si measurements, such studies may prove impractical.