A control strategy for fertiliser granulation circuits is developed by employing a dynamic model which is based on a sound understanding of granulation mechanisms. The overall study of the control strategies includes selection of the controlled variables, determination of the control structure and case studies. Recycle size distribution is chosen as an easily measured controlled variable and a PI controller is implemented in the process. An examination for the control of a di-ammonium phosphate granulation process is carried out by using the NIMBUS Simulator. The simulation results show the off-set of the granule mean size and the moisture content caused by the step change disturbance is reduced greatly compared with the open loop simulation. The issues for the control of the large time lag and the large recycle stream are discussed.