The aim of this paper is develop a nonlinear optimisation model for the determination of the optimised water allocation and cropping pattern under adequate and limited water supplies. The water productivity index defined as the net profit to the volume of water used was considered as the objective function. The proposed model was executed for the Ghazvin Irrigation Network located in a semi-arid region in Iran. The results showed that among the crop types grown in the region, onion and alfalfa have the highest and lowest water productivity value, respectively. These values under drought conditions for the optimal cropping pattern of the two crops were estimated at 75 068.86 and 3054.18 Rls m À3 . The findings indicated that the overall water productivity of the irrigation network with relevant cropping pattern management might be raised to as high as 12 700 Rls m À3 under drought conditions. In normal and wet years, depending on the water available and the optimal cropping pattern, the values for this index were estimated to be 15 600 and 12 900 Rls m À3 , respectively. For the existing cropping pattern, overall irrigation network productivity is estimated at 10 600 Rls m À3 . Hence, the results demonstrated that the water productivity of an irrigation network could be improved as result of the optimal cropping pattern and deficit irrigation. For the study area, the maximum variations of this index may be fixed around 18% for different water regimes. The evaluations emphasise the important role of optimisation models in improving irrigation network efficiency and managing water in a sustainable manner. Copyright # 2008 John Wiley & Sons, Ltd.