A coupled ice-ocean Arctic ocean model is forced with output from a Canadian Climate Center (CCC) model of the atmosphere. Two cases are considered, the first where the atmospheric CO2 is ''normal'' and the second where the atmospheric CO2 is ''doubled.'' The CCC normal forcing data are also compared with observed fields. It is found that the forcing fields from the CCC normal model run are quite similar to observed values of surface temperature and humidity. The wind stresses are, however, lower than observed geostrophic values and are hence scaled to more closely match observed values. Significant differences are found between the normal and doubled CO2 forcing fields in regard to temperature and humidity forcing fields. In particular, the surface air temperature is typically 10¿ Celsius warmer in the doubled case for the winter months, and the relative humidity is up to 50% higher in the summer months. The surface moisture flux from the CCC data is found to be relatively insensitive to doubling of CO2, with more precipitation less evaporation (P--E) in winter months. When the ice-ocean model is run, less ice volume and increased areas of open ocean are observed with the doubled forcing, however, the decrease in ice volume is not as large as previous, simpler models have suggested and the model ice cover differs significantly from the ice cover from the atmospheric model. It is also found that there are differences in the oceanic temperature and salinity fields consistent with the forcing inputs, but these appear to be minimized by the insulating properties of the ice cover. On balance, the Arctic ice field appears to act as a regulator of climate change, rather than an accelerator. ¿ American Geophysical Union 1995