A suite of the historical simulations run with the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4) models forced by greenhouse gases, aerosols, stratospheric ozone depletion, and volcanic eruptions and a second suite of simulations forced by increasing CO(2) concentrations alone are compared with observations for the reference interval 1965-2000. Surface air temperature trends are disaggregated by boreal cold (November-April) versus warm (May-October) seasons and by high latitude northern (N: 40°-90 °N) versus southern (S: 60 °S-40 °N) domains. A dynamical adjustment is applied to remove the component of the cold-season surface air temperature trends (over land areas poleward of 40 °N) that are attributable to changing atmospheric circulation patterns. The model simulations do not simulate the full extent of the wintertime warming over the high-latitude Northern Hemisphere continents during the later 20th century, much of which was dynamically induced. Expressed as fractions of the concurrent trend in global-mean sea surface temperature, the relative magnitude of the dynamically induced wintertime warming over domain N in the observations, the simulations with multiple forcings, and the runs forced by the buildup of greenhouse gases only is 721, and roughly comparable to the relative magnitude of the concurrent sea-level pressure trends. These results support the notion that the enhanced wintertime warming over high northern latitudes from 1965 to 2000 was mainly a reflection of unforced variability of the coupled climate system. Some of the simulations exhibit an enhancement of the warming along the Arctic coast, suggestive of exaggerated feedbacks.