The variability of surface temperature simulated by a global climate model with a simple mixed-layer ocean is analyzed. The simulated diurnal and seasonal ranges of temperature are compared with observation, as is the day-to-day and interannual variability of temperature. The qualitative changes in these quantities due to doubling atmospheric carbon dioxide concentration are also presented. The simulation of the seasonal cycle of surface temperature has a cold bias in much of the extratropics, including central Europe, even allowing for the difficulties in comparing grid-box surface temperatures with station temperature at screen height. The simulated diurnal range of temperature for present-day climate is similar to that observed, though the diurnal cycle in the model in midlatitudes is generally less than observed. On doubling C02, the diurnal range over land decrease by 0.3°C whereas mean temperatures increase by 6.3°C (global average over land). In CO2-doubling experiments with a one-dimensio... Abstract The variability of surface temperature simulated by a global climate model with a simple mixed-layer ocean is analyzed. The simulated diurnal and seasonal ranges of temperature are compared with observation, as is the day-to-day and interannual variability of temperature. The qualitative changes in these quantities due to doubling atmospheric carbon dioxide concentration are also presented. The simulation of the seasonal cycle of surface temperature has a cold bias in much of the extratropics, including central Europe, even allowing for the difficulties in comparing grid-box surface temperatures with station temperature at screen height. The simulated diurnal range of temperature for present-day climate is similar to that observed, though the diurnal cycle in the model in midlatitudes is generally less than observed. On doubling C02, the diurnal range over land decrease by 0.3°C whereas mean temperatures increase by 6.3°C (global average over land). In CO2-doubling experiments with a one-dimensio...