The dynamic and stochastic components of a state model are identified fora real time multivariable pressure tank process. Some simplifying assumptions for the stochastic components of the model allow for easy determination of the Kalman filter and subsequent on-line implementation of a Linear Quadratic control law. The controller behaves well for both load and set point disturbances for this process which is strongly interactive and nonlinear. Some practical suggestions for real time implementation are given.