Multivariable process control problems contain many difficult challenges for both statisticians and control engineers. The approach presented in this paper includes a number of novel features that resolve some troublesome practical problems in process control applications. These contributions and the success of the control algorithm for the reactor example described in the paper should encourage statisticians to become more closely involved in the important area of process control. Unfortunately some confusion exists in the descriptions of the control objectives and the chemical reactions in the example with the result that the example is not as effective as it might have been in illustrating the significant statistical contributions ofthe paper.