To predict and control the silicon content of the hot metal ([%Si]) in blast furnace ironmaking process, an expert network consisted of a neural network model and an expert system has been established and tested off-line with the practical process data. The applicability of either the neural network model or the expert system for the prediction of [%Si] is determined by the experiential rules. In general, the predictions are made by the neural network model in the periods of the normal operation of the furnace, and performed by the expert system in the periods of the unsteady operation of the furnace. The operation guidance for adapting the furnace process is recommended by the expert system