Climate change simulations made with coupled global climate models typically show a marked hemispheric asymmetry with more warming in the northern high latitudes than in the south. This asymmetry is ascribed to heat uptake by the ocean at high southern latitudes. A recent version of the CCCma climate model exhibits a much more symmetric warming, compared to an earlier version, and agrees somewhat better with observed 20th century trends. This is associated with an improved parameterization of ocean mixing which results in a decrease in heat penetration into the Southern Ocean, in accord with earlier ocean-only and simple coupled model investigations. The global average warming and the net penetration of heat into the global ocean (and hence its thermal expansion) are essentially unchanged. Observed trends in sea-ice extent over the past two decades exhibit hemispheric asymmetry with a statistically significant decrease in northern but not in southern ice cover. Both model versions are consistent with these observations implying that observed ice extent is not yet an indicator of asymmetry in future global warming. Taken together, these results suggest that southern hemisphere climate warming at a rate comparable to that in the northern hemisphere should be considered a realistic possiblity. ¿ 2001 American Geophysical Union