This study concerns analysis of climatic variability in general circulation models, first to determine how faithfully the models reproduce measures of the present variability, and then to examine how variability changes in CO2 perturbed cases. The control run output of three different versions of the NCAR Community Climate Model is examined for selected regions of the United States including the Great Plains, the Southeast, the Greak Lakes region, and the upper West Coast. Mainly temperature and precipitation are analyzed, but some preliminary analysis of relative humidity is also included. Interannual variability is analyzed in one version (Chervin) for only the present day climate. Results indicate that this version underestimates the interannual variability of temperature, but generally successfully reproduces the relative variability of precipitation. Two versions of the CCM tend to overestimate daily temperature variability while the third (Dickinson) accurately reproduces or underestimates it. Reasons for these differences are discussed. Daily precipitation variability is overestimated by two model versions (Chervin and Dickinson) but fairly well reproduced by the Washington version. The CO2 perturbed run of one version (Washington) produces mixed results for changes in daily temperature variability and tends to produce increased daily variability for precipitation. ¿ American Geophysical Union 1990