We present the results of a sensitivity study involving modifications to the simple land surface scheme implemented in the second-generation atmospheric general circulation model (GCMII) of the Canadian Climate Centre for Modelling and Analysis (CCCma), under paleoclimatic boundary conditions characteristic of 6000 calendar years before present (6 ka BP). The land surface parameterization is modified in two primary respects. Firstly, we modify the space dependant bucket depth scheme in the original model to one in which this depth is taken to be constant. Secondly, we modify the evapotranspiration parameterization from the space dependant form employed in the control model to a more conventional space independent scheme. In all, 4 experiments have been performed to enable us to resolve both the modern control and the 6 ka BP response to the land surface modifications. A subset of these simulations is also compared with results obtained using other models in the context of the Paleoclimate Model Intercomparison Project (PMIP) to investigate the mid-latitude behaviour of these models to reveal the extent to which model response to a change in radiative forcing may be significantly influenced by changes in the land surface parameterization. These comparisons reveal that the original Canadian GCMII model is an extreme outlier among the members of the set of all models in that its Northern Hemisphere mid-latitude surface continental response to the 6 ka BP insolation anomaly is significantly cold biased in the summer season. We investigate the extent to which this anomalous behaviour may be explained as a consequence of modifications to the land surface parameterizations employed in GCMII. Our results reveal a strong sensitivity in the modern control model to changes in bucket depth, but not to the modification of the evapotranspiration scheme. More interesting, however, is the fact that the model climate sensitivity at 6 ka BP is influenced both by changes in bucket depth and by changes in the evapotranspiration scheme. A detailed investigation of the surface energy balance, cloud cover, surface albedo, and snow cover reveals the role of each of the components that contribute to the 6 ka BP surface temperature response. Comparison of the predictions of the CCCma model with proxy climate indicators of lake levels and surface temperature over Canada, furthermore, demonstrate the extreme sensitivity of climate predictions for this geographical region to changes in the manner in which land surface processes are represented.