Prediction of nitrogen (N) mineralization is important for specifying the optimum rate of N fertilizer for flooded rice at the time of sowing. To develop a predictive test, soils (0'0.1 m) were sampled from 22 farms throughout the rice-growing region of southern Australia over a 4-year period. Near infrared reflectance (NIR) spectra of the soils were compared with sixteen biological and chemical soil tests for the prediction of N-uptake by rice plants from these soils in the field and glasshouse. The aim of the study was to develop a soil-NIR calibration as an accurate, rapid and economical mineralization test. Nitrogen uptake by field-grown and glasshouse-grown plants was poorly correlated (r = 0.30), even though significant NIR calibrations were developed with both. Since N uptake by rice in the field was affected by varying weather and management, the field calibration is probably spurious. The calibration of soil NIR spectra with N uptake by glasshouse plants was satisfactory, with a standard error (SE) of 13 kg ha'1 over a range of 11 ' 95 kg ha'1, and a correlation between calculated and measured N uptake (r = 0.87, P<0.001). An even better soil-NIR calibration was found with N-mineralization after 21 days of anaerobic incubation (SE 16 mg kg'1, range 52'175 mg kg'1). Analysis of the soil spectra showed that similar wavelengths were correlated with both plant-N uptake and mineralization. NIR spectroscopy shows considerable potential to predict soil N mineralization, and may assist future fertiliser decision support.