Crop recovery of nitrogen (N) fertiliser in flooded rice systems is low relative to fertiliser N recoveries in aerobic crops, and the N losses have environmental consequences. Recent water shortages across the globe have seen a move towards alternative water management strategies such as delayed permanent water (DPW, also known as delayed flood). To investigate whether N fertiliser regimes used in DPW systems result in greater recovery of N fertiliser than traditional continuously flooded (CF) rice systems, we conducted a multi-N rate field trial using 15N-labelled urea. Around 27% of the 15N-labelled fertiliser was recovered in aboveground biomass at maturity, regardless of water regime or N fertiliser rate, and approximately 20% recovered in the soil to 300 mm depth. Plants in the CF system accumulated more total N at each rate of applied N fertiliser than plants in the DPW system due to greater exploitation of native soil N reserves, presumably because the earlier application of N fertiliser in the CF systems led to greater early growth and higher crop N demand. The greater crop biomass production as a result of higher N uptake in the CF system did not increase grain yields above those observed in the DPW system, likely due to cold weather damage. In the following season at the same site, a single N rate (150 kg N ha−1) trial found no significant differences in crop N uptake, biomass yields, grain yields or 15N-labelled urea recovery in DPW, CF and drill sown-CF (DS-CF) treatments. However, owing to higher 15N fertiliser recovery in the 0–100 mm soil horizon, total plant + soil recovery of 15N was significantly higher in the CF treatment (63%) than the DS-CF and DPW treatments (around 50% recoveries). The loss of 40–50% of the applied N (presumably as NH3 or N2) in both seasons regardless of watering regime suggests that new fertiliser N management strategies beyond optimising the rate and timing of urea application are needed, particularly in light of increasing N fertiliser prices.