A global, two-dimensional, one-level seasonal energy balance model (linear in temperature) is asynchronously coupled to vertically integrated ice-flow models (which depend on latitude and longitude) in order to study the response of the atmosphere-ocean-cryosphere-lithosphere system to variable solar forcing during the last ice age cycle of late Pleistocene time. At the last glacial maximum, the model is shown to satisfactorily simulate both the geographic distribution of ice volume and the individual ice sheet masses within the North American and European ice sheet complexes as they have recently been inferred on the basis of relative sea level data (Tushingham and Peltier, 1991). Because the energy balance model employed here does not take mean ocean currents into account, the absence of the moderating influence of the warm currents in the North Atlantic Ocean leads to a simulation such that surface temperatures are too cold over Scandinavia and the Barents and Kara seas, which leads to an erroneous European ice sheet even in the interglacial phase of the glacial-interglacial cycle. The sensitivity of the coupled model to variable solar forcing is also insufficient to explain the terminations that are such a striking feature of individual ice age cycles (e.g., Broecker and van Donk, 1970), and this suggests that further physical mechanisms will have to be added to the model in order to explain this important phenomenon. Suggestions are provided as to what these missing physical ingredients might be. ¿1991 American Geophysical Union