Model systems such as SUNDIAL provide arable N fertiliser advice by modelling the supply from soil, crop uptake, and losses of N. However, poor knowledge of future weather reduces accuracy. Improvements in forecasting may lead to greater use of forecasts within the food chain. We attempt to quantify the potential of medium-term weather forecasts to improve N fertiliser recommendations on arable crops in England and Wales. We used a weather generator to produce weekly weather variables. The generator was modified to produce paired sets of simulated actual weather and forecasted weather at a range of forecast accuracies and durations, for 10 regions representing England and Wales. SUNDIAL was used to test the effect of prior knowledge of weather following the date of N fertiliser application, for four different crops. Changes due to forecast quality were calculated, in crop N uptake, N leaching and denitrification. Yield and gross profit changes were estimated from N uptake. Accurate forecasts reduced the risk of under-application of N. Three-week forecasts increased crop uptake by an average of 2 kg N ha^-1, and profit by £23M per annum. The national decrease in losses was small. Similar improvements may be expected for other dynamic recommendation systems that exploit post-application weather.