Molten metal carbon is an important parameter to be controlled during the BOF operation. In this paper, a fuzzy logic system for carbon sensing in a basic oxygen furnace (BOF) is presented. This objective system is identified from input-output data of the system by applying the subtractive clustering algorithm. The input data presents the optical radiations emitted from the BOF mouth and the output presents the carbon content of the molten metal contained in BOF in order to provide an efficient decision support application. The variables employed to develop the fuzzy logic system are obtained through an optical sensor located adjacent a full scale BOF at Nanjing steels’. Preliminary results are really encouraging.