Abstract: A new framework was presented for the accurate modeling and prediction of the reblown oxygen and the added coolant in dynamic basic-oxygen-furnace (EOF) steelmaking processes. The proposed method takes advantages of the modeling approach based on mechanism and uses adaptive neural-network-fuzzy-inference system (ANFIS) to compensate for the BOF modeling uncertainties based on mechanism. In the ANFIS compensating model, the first-order Takagi-Sugeno type fuzzy rules were employed and a hybrid algorithm combining the least square method (LSM) and the gradient descent method was adopted to obtain the model structure. The practical data of a 180 t converter were simulated. The simulated results are close to the practical values. The method is practicable and effective.