This paper statistically analyzes and compares the variability of the observed and simulated (by the Geophysical Fluid Dynamic Laboratory general circulation model (GCM)) northern hemisphere monthly surface air temperatures, averaged within different latitude bands. The objects used for the analysis are: seasonal cycle of variances and univariate spectral and autocorrelation functions of zonal time series. The estimates reveal a significant latitudinal trend of all second moments. A comparison shows a statistically significant distinction of the variances of the observed and simulated stochastic systems. In some cases the shapes of the corresponding univariate spectra and autocorrelation functions are quite similar, but their numerical values and some features differ markedly, especially for the tropical regions. The approach developed gives reasonable descriptions of the univariate temperature processes, thereby providing an opportunity for solving some of the vital problems of theoretical and practical aspects surrounding validation and diagnosis of GCMs. ¿ 1997 American Geophysical Union