Abstract: A quasi–operational experiment for 30–day forecast has been made for morthly anomalies of 100 hPa, 500 hPa, SLP and SLT during January through December of 1988 from a nonsteady atmosphere—earth surface coupled anomaly model. The results show that the forecast of tendency of 100 hPa monthly anomaly is best and that of SLP is not too good, the model forecasts for both 500 hPa and surface temperature monthly anomalies are better than persistence forecasts. It means that this model can be forecast anomaly field in large scale. The results also show that it is better in the low and mid–latitude regions than in the mid–high latitudes. But the forecast effects are not quite the same for the different levels and months and the forecasts for low pressure systems are very satisfactory. The model has a defect in the spatial resolution, which is too smooth. So, it needs more experiments and improvements.