The papermaking process can be viewed as a time-varying stochastic system. The purpose of the control is to minimize the variance of moisture and basis weight of the manufactured paper. In the framework of stochastic theory and recently published papers on adaptive control an algorithm has been developed to cover the abovementioned purpose. The strategy which hardly requires any a priori knowledge about the process is, in fact, a selfadjusting regulator. The central part of the algorithm is a compact identification scheme which gives directly the optimal control action. It is notable that contrary to normal identification and control methods the process dynamics never need to be calculated. The algorithm has been applied to several real-time computer control systems for paper machines. Up to now medium-scale computers have been used, but as the control program is powerful and short, some hundred words in length, it is in fact sufficient to perform the control with a mini-computer. Before the implementation of the algorithm for the above processes, the control system was built around digitalized PIcontrollers, the parameters of which were carefully chosen and constant. A great number of comparative tests have been made in order to evaluate the efficiency of the adaptive regulator. The results were so successful that the algorithm has been permanently installed in the systems. The algorithm has not only decreased the output variance but also reduced the losses at quality changes and mill set-ups. Finally, it should be mentioned that the general nature of the algorithm permits application to many other types of processes.