We simulate climate change for the 2-year period following the eruption of Mount Pinatubo in the Philippines on June 15, 1991, with the ECHAM4 general circulation model (GCM). The model was forced by realistic aerosol spatial-time distributions and spectral radiative characteristics calculated using Stratospheric Aerosol and Gas Experiment II extinctions and Upper Atmosphere Research Satellite-retrieved effective radii. We calculate statistical ensembles of GCM simulations with and without volcanic aerosols for 2 years after the eruption for three different sea surface temperatures (SSTs): climatological SST, El Ni¿o-type SST of 1991--1993, and La Ni¿a-type SST of 1984--1986. We performed detailed comparisons of calculated fields with observations. We analyzed the atmospheric response to Pinatubo radiative forcing and the ability of the GCM to reproduce it with different SSTs. The temperature of the tropical lower stratosphere increased by 4 K because of aerosol absorption of terrestrial longwave and solar near-infrared radiation. The heating is larger than observed, but that is because in this simulation we did not account for quasi-biennial oscillation (QBO) cooling and the cooling effects of volcanically induced ozone depletion. We estimated that both QBO and ozone depletion decrease the stratospheric temperature by about 2 K. The remaining 2 K stratospheric warming is in good agreement with observations. By comparing the runs with the Pinatubo aerosol forcing with those with no aerosols, we find that the model calculates a general cooling of the global troposphere, but with a clear winter warming pattern of surface air temperature over Northern Hemisphere continents. This pattern is consistent with the observed temperature patterns. The stratospheric heating and tropospheric summer cooling are directly caused by aerosol radiative effects, but the winter warming is indirect, produced by dynamical responses to the enhanced stratospheric latitudinal temperature gradient. The aerosol radiative forcing, stratospheric thermal response, and summer tropospheric cooling do not depend significantly on SST. The stratosphere-troposphere dynamic interactions and tropospheric climate response in winter are sensitive to SST. ¿ 1999 American Geophysical Union