We have measured 230Th-normalized opal fluxes in several cores from the eastern equatorial Pacific (EEP) to test the validity of the "silica leakage" hypothesis, which purports that redistribution of silicic acid from the Southern Ocean to the low latitudes was responsible for a significant portion of the reduction in atmospheric carbon dioxide (CO2) during the last glacial period. The silica leakage hypothesis predicts higher opal fluxes in the EEP and lower opal fluxes in the Southern Ocean during periods of low atmospheric CO2. These predictions are not borne by the sedimentary record during glacial oxygen isotope stage 2 (OIS 2, 13--27 kyrs B.P.). However, we find a prominent opal flux maximum in the EEP in the middle of OIS 3 (ca. 40--60 kyrs BP) coinciding with low opal fluxes in several cores from the subantarctic zone. This observation is consistent with silica leakage from the Southern Ocean to the equatorial upwelling region during OIS 3, when both low dust flux and extended sea ice could have contributed to limiting diatom productivity in the Southern Ocean. Since this event is not associated with a clear minimum in the Vostok ice record of CO2, its impact on atmospheric CO2 appears to be small.