Maps of upper-level and surface winds and of surface temperature and precipitation illustrate the results of a sequence of global paleoclimatic simulations spanning the past 21,000yr for North America. We review (a) the large-scale features of circulation, temperature, and precipitation that appear in the simulations from the NCAR Community Climate Model Version 1 (CCM 1), (b) the implications of the simulated climate for the past continental-scale distributions of three plant taxa (Picea spp., Pseudotsuga menziesii, and Artemisia tridentata), which are broadly representative of the vegetation across the continent, and (c) the potential explanations in terms of atmospheric circulation or surface energy- and water-balance processes for mismatches between the simulations and observations. Most of the broad-scale features of previous paleoclimatic simulations with the NCAR CCM 0 for North America are present in the current simulations. Many of the elements of a conceptual model (based on previous climate simulations) that describes the controls of paleoclimatic variations across North America during the past 21,000yr are found in simulations reviewed here. These include (1) displacement of the jet stream by the Laurentide Ice Sheet to the south of its present position in both winter and summer, (2) generation of a ‘glacial anticyclone’ over the ice sheet at the LGM, and the consequent induction of large-scale sinking motions induced over eastern North America, (3) changes in the strength of surface atmospheric circulation features through time, including weakening of the Aleutian low in winter, and strengthening of the eastern Pacific and Bermuda high-pressure systems in summer as the ice sheet decreased in size, (4) development of a ‘heat low’ at the surface and a strengthened ridge in the upper-atmosphere over the continent at the time of the maximum summer insolation anomaly, (5) increases in summer temperature earlier in regions remote from the ice sheet (these increases appear earlier in the present (CCM 1) simulations than in the previous (CCM 0) ones, however), and (6) continuation of negative winter temperature anomalies into the middle Holocene. In general, simulated surface conditions that are discordant with paleoenvironmental observations can be attributed to the simulation of particular atmospheric circulation patterns (e.g. those that suppress precipitation or advect warm air into a region), with these mismatches amplified in Beringia and the southeastern United States by surface energy- and water-balance processes.