Abstract This paper presents an Artificial Neural Network (ANN) based adaptive control scheme for a multistage flash (MSF) sea water desalination plant. The detailed dynamic model of an 18-stage MSF plant has been developed and validated by measurements on the actual plant which is in operation in the United Arab Emirates. The non-linear behaviour of the model makes a fixed controller unfit for operation in the operating region that is normally chosen in which the controller parameters have to be tuned at each of a number of points based on the linearized version of the plant model. An ANN is trained with the information derived from simulation based optimal design at selected points in the operating region. The underlying control law is one which uses integral performance criteria and provides optimal PID controller parameters. With the ANN so trained and inserted in an adaptive control loop, the system maintains optimality in the entire operating region.