Management strategies that improve water efficiency in water-limited rice systems are needed for sustainable production. In southeast Australia growers are increasing implementing drill seeding and also delayed permanent water (DPW) irrigation practice to improve water productivity. This change in timing of permanent water application has a large influence on crop phenology which impacts the timing of crop management practices. Two types of phenological models were assessed to predict panicle initiation (PI) timing in fields managed using drill sowing and DPW. A single-stage model was contrasted with a two-stage efficiency model. The single-stage model assumed temperature across the planting to PI period equally contributes to PI timing. The two-stage efficiency model allowed for differential temperature efficiencies between the pre (aerobic) and post (anaerobic) permanent water periods. Four temperature indices, one growing degree day and three parameterisations of degree day (DD) were tested. Observations of PI from seven seasons and seven locations were used to parameterise (n = 55) and validate (n = 7) the models. The best model was for the two-stage efficiency approach using the original DD parameters with RMSE of 3.8 and 4.4 days for the parameterising and validating data, respectively. The methodology outlined can be used for other varieties, physiological stages and water management strategies to develop models to better predict phenology in rice systems managed with DPW.