The feasibility of long-range weather prediction through the use of global general circulation models (GCMs) was investigated. A climate model was developed to simulate the monthly mean state of the atmosphere from real global initial data at the beginning of the month. The model contains the same dynamic and physical ingredients as most numerical weather prediction models and GCMs. The model generates a one-day global simulation on the 8 x 10 grid in four minutes (on an IBM 360/95 computer), so that a 30 day forecast can be executed in two hours. The high speed of the model is achieved mainly at the price of its coarse resolution, which requires certain parameterizations of surface boundary conditions, as well as inherent filtering of smaller-scale features of the initial state. Spar, J. Unspecified Center NASA-CR-157893 NGR-33-016-086