A thermodynamically motivated method for the identification of dominant; variables used in partial control is applied to the Tennessee Eastman challenge process. The method shows that the reactor temperature together with the vapor-phase composition of component A are dominant variables. It is also shown that the temperature in the separator is dominant for this unit and that the overhead vapor composition is a dominant variable for the stripper. When a complete partial control structure is formed by feedback control of all the dominant variables, it turns out that an insufficient number of manipulated variables remain to satisfy inventory and component balances. A reduced partial control structure is thus suggested. It is demonstrated that this design is superior to other control solutions to the Eastman process in that it can easily attain and hold the plant at its maximum production rate.