A coarse-grid global ocean general circulation model (OGCM) is used to determine the role of sub-grid scale eddy parametrization schemes in the response to idealized changes in the surface heat flux, of the same order as expected under increased atmospheric CO2 concentrations. Two schemes are employed. The first (H) incorporates standard horizontal mixing, whereas the second (G) combines both enhanced isopycnal mixing and eddy-induced transport. Uniform surface heating anomalies of +2 W m-2 and −2 W m-2 are applied for 50 years, and the results are compared with a control experiment in which no anomalous heating is imposed. A passive “heat” tracer is applied uniformly (at a rate of 2 W m-2 for 50 years) in a separate experiment. The sea-surface temperature response to global surface heating is generally larger in G, especially in the northern subtropical gyres, along the southern coast of Australia and off the Antarctic coast. A pronounced interhemispheric asymmetry (primarily arising from an anomalous response south of 35 °S) is evident in both H and G. The surface trapping of passive tracers in the Southern Hemisphere is generally greater in G than it is in H, and is particularly pronounced along the prime meridian (0 °E). Dynamical changes (i.e., changes in horizontal and vertical currents, convection, and preferred mixing and eddy transport pathways) enhance surface warming in the tropics and subtropics in both G and H. They are dominated by an anomalous meridional overturning centred on the equator, which may also operate in greenhouse warming experiments using coupled atmosphere-ocean GCMs. Over the Southern Ocean the passive tracer experiments and associated ventilation rates suggest that surface warming will be greater in G than in H. In fact, the contrast between the dynamical responses evident in G and H in the actual heating experiments leads to a situation in which the reverse is often true. Overall, dynamical changes enhance the interhemispheric assymetry, more so in G than in H.