Quadruple tank process is a bench mark system to demonstrate Multi Input Multi Output (MIMO) process with non minimum phase behaviour. Controlling of such MOMO non minimum phase system is difficult using conventional methods. Model based control will be a good choice. Mathematical modelling may not be suitable for real time control because of it limitations due to assumptions in deriving the model. Modelling the system using experimental data will be an ideal choice. Neural network modelling is one of the methods using input output data. In this work, an attempt was made to model the process using neural network. The results are encouraging.