Variable-rate irrigation technology can reduce water use in centre pivot and lateral move irrigation systems through application of irrigation according to spatially varied soil-water profiles. However, filling the profile may not maximise yield because of variations in crop response and water requirements with crop stage. For example, cotton crops produce optimal yield under slight water stress during early stages. An irrigation strategy ‘Model Predictive Control’ has been implemented that accounts for changes in crop water requirements at different growth stages using biophysical crop models. This strategy involves automatically and iteratively executing the biophysical crop model APSIM parameterised with local soil and weather information, with different irrigation depths, to identify which combination maximises yield with the minimum depth of water application. This strategy has potential to address spatial and temporal variations in crop water requirements but has not previously been evaluated for variable-rate irrigation in the field. This paper reports field trials conducted over four cotton (Gossypium hirsutum L.) seasons and two perennial ryegrass (Lolium. perenne L.) seasons to evaluate the accuracy of the yield prediction of the biophysical model and compare field performance of irrigation strategies: uniform irrigation and variable-rate irrigation using a fixed underlying map, soil-water sensors and Model Predictive Control. Yield was most accurately predicted using on-site weather data and field soil core information, with R² = 0.733 and RMSE = 153.9 kg/ha for cotton, and R² = 0.336 and RMSE = 295.3 kg/ha for ryegrass. For cotton, Model Predictive Control led to 4.9% more yield with 5.6% reduction in water application, mainly through reduced water after peak bloom and/or open boll physiological stages. For grazed ryegrass, the Model Predictive Control strategy led to 8.5% more yield with 5.4% reduction in water application, potentially caused by reduced applications after grazing events. Further work includes evaluating the Model Predictive Control strategy with control of irrigation event timing under a broader range of field conditions to identify parameters to provide greatest economic return, and to refine biophysical models for improved performance of optimisation in the strategy.