Geographically based calculations for 18,000 years ago (18 ka) and today were made to examine the potential effects on terrestrial chemical erosion of changes in lithology and hydrology on glacial-interglacial timescales. Runoff fields were derived from general circulation model predictions of precipitation minus evaporation. Then global lithologic maps were prepared, so that empirical relationships for runoff versus bicarbonate flux for different rock types could be used to calculate global chemical erosion rates. Assuming that significant chemical erosion does not occur beneath ice sheets, we find that chemical erosion in ice-free areas at 18 ka was only slightly greater than today. This result arises because the amount of land covered by ice sheets is roughly compensated for by exposed shelf areas and because there is little difference in global runoff. The small (∼20%) increase in the global chemical erosion rate during glacial conditions is due to exposure on the shelves of a relatively high proportion of carbonates (which weather faster than average). Data on glacial/interglacial movements of the calcite compensation depth (CCD) are inconclusive but seem to indicate little change, whereas even a 20% increase in the riverine bicarbonate flux should cause an observable deepening. If the CCD did not indeed change, then our predicted increase in the bicarbonate flux from land during glaciations would have to be accommodated by other means, such as increased carbonate productivity. About half of the observed decrease in atmospheric pCO2 at 18 ka could be explained if increased silicate chemical erosion accompanied increased total chemical erosion in ice-free areas. Consideration of the maximum possible effect of meltwater at ice margins leads to an 80% increase in the global chemical erosion rate at 18 ka. Such an increase would lead to a larger and faster drop in atmospheric pCO2 but also to an excessive deepening of the global CCD; this scenario seems unrealistic. Our results are contrary to interpretations of a much higher silicate chemical erosion rate during glaciations based on recently published records of Quaternary marine Sr isotopic ratios. Therefore, if these records are indeed globally representative and entirely due to increased silicate chemical erosion (as opposed to changes in source areas or hydrothermal inputs), then other factors, such as a shift in the global weathering regime which we do not explicitly consider here, may be involved. Sensitivity tests show that considering spatial heterogeneities in lithology and runoff leads to lower predictions of global chemical erosion rates than what one would obtain by considering only global averages in a non-spatially-resolved calculation.