It is established that an algorithm for parametric methods of pattern recognition can be used to build a classifying rule that allows to identify a section of the duplex-melting process «cupola – electric arc furnace», which has a lower index of parametric reliability. It is shown that in this case it is sufficient to build a linear discriminant function.The resulting classifying rule provides a high recognition rate (91.4 % and 97.1 %) of object belonging to class A (melting in cupola) and class B (melting in electric arc furnace), respectively. This allows it to be recommended for decision support systems for desulphurization and dephosphorization control in the duplex-melting process «cupola – electric arc furnace». The application of this rule opens the possibility of improving the quality of desulphurization and dephosphorization control due to the correct choice of control actions.Additional opportunities for using the above results in industrial conditions are associated with the use of the classifying rule in decision support information systems. It can be implemented by integration of the appropriate mathematical description into the melting information control system (ICS).