This paper reports the performance evaluations undertaken on a real-time optimisation system for furrow irrigation (AutoFurrow). Trials for the system were undertaken on commercial furrow-irrigated cotton properties near St George and Dalby, Queensland Australia. The system performed robustly in the field and demonstrated its potential for substantial water savings; however, the results suggested that there was further scope for improvement in performance. To identify opportunities for improvement, the surface irrigation simulation model SISCO was used to investigate the effect of varying: the objective function, flow rate, irrigation deficit, infiltration scaling process and the model infiltration curve. It was found that a simple objective function that aims to maximise application efficiency (AE) can deliver accurate prediction of the irrigation performance and potentially add to the robustness of the optimisation process. It was also demonstrated that if a suitable flow rate is selected initially, then no further change is warranted. The predicted time to cut-off (TCO) was relatively insensitive to the irrigation deficit; however, any change in the irrigation deficit altered the AE predicted.