Productivity and response to nitrogen (N) fertilizer were measured on 84 irrigated rice crops growing over two seasons on farms in the Riverina region of south-eastern Australia. The results were correlated with 18 tests of the N status of the top 10cm of soil to develop a method to specify optimum N fertilizer application before sowing. In this environment, nitrogen fertilizer applied before sowing produces higher yields than the existing pre-sowing-topdressing split, provided the application rate is not so high that it leads to cold damage or lodging. Yield, biomass and N uptake varied greatly between the 84 crops, but the mean apparent N recovery of 50–60% and the N-use efficiency of 22kggrainkgN −1 were high by commercial standards. Of the 18 soil-N tests, the one most closely correlated to crop productivity used anaerobic incubation for 21 days at 40°C. The standard error of this test for predicting crop biomass was at least 74kgNha −1 , which is double the standard error of the current plant test and is unacceptably high for providing recommendations. A comparison of commercial N-application rates and the economic optimum rates showed that ricegrowers tended to overfertilize crops growing on the most fertile fields and underfertilize crops growing on the least fertile fields. Suggested strategies to increase N-use efficiency are to improve the accuracy of the soil test by considering factors such as the status of other nutrients, to inform ricegrowers of their tendency to overfertilize high-yielding crops and underfertilize low-yielding crops, and use a soil test only to discourage overfertilization.