The permeability index for blast furnaces is an important monitoring parameter in their operation. Proper trend prediction of the permeability index is important for good operation. Support vector machines (SVM) combined with wavelet analysis are adopted to build a forecasting model. Four historic values of a permeability index are analyzed by a wavelet analysis via seven levels. Based on eight wavelet analyzed values and combined with operating parameters, eight submodels are built using the least square support vector machines method. The prediction components are reconstructed to obtain a forecast. The details of modeling, validation and result analyses are presented.