Two contrasting hypotheses have recently been proposed for the past long-term relation between atmospheric CO2 and the carbonate-silicate geochemical cycle. One approach (Berner, 1990) suggests that CO2 levels have varied in a manner that has maintained chemical weathering and carbonate sedimentation at a steady state with respect to tectonically controlled decarbonation reactions. A second approach (Raymo et al., 1988), applied specificlly to the late Cenozoic, suggests a decrease in CO2 caused by an uplift-induced increase in chemical weathering, without regard to the rate of decarbonation. According to the steady-state (first) hypothesis, increased weathering and carbonate sedimentation are generally associated with increasing atmospheric CO2, whereas the uplift (second) hypothesis implies decreasing CO2 under the same conditions. An ocean-atmosphere-sediment model has been used to assess the response of atmospheric CO2 and carbonate sedimentation to global perturbations in chemical weathering and decarbonation reactions. Although this assessment is theoretical and cannot yet be related to the geologic record, the model simulations compare steady-state and non-steady-state carbonate-silicate cycle response. The e-fold response time of the 'CO2-weathering' feedback mechanism is between 300 and 400 ka. The response of carbonate sedimentation is much more rapid. These response times provide a measure of the strength of steady-state assumptions, and imply that certain systematic relations are sustained throughout steady-state and non-steady-state scenarios for the carbonate-silicate cycle. The simulations suggest that feedbacks can maintain the system near a steady state, but that non-steady-state effects may contribute to long-term trends. The steady-state and uplift hypotheses are not necessarily incompatible over time scales of a few million years. ?? 1991.