Abstract: In the field of crop nutrition diagnosis by spectrum technology, it is greatly important that selecting right spectrum parameters to improve accuracy and precision. The objectives of this experiment were to identify wavelengths and/or their combinations that are indicative of nitrogen nutritional condition and to analyze the accuracy of different forms spectral parameters for nitrogen nutrition diagnosis. Corn (Zea may L.) different layer leaf reflectance spectra and nitrogen content of different nitrogen treatments were measured at key development stages. Correlation analysis between spectral reflectance of different layer leaf and nitrogen content were made and linear regression equations were constructed between spectral parameters and nitrogen content. The accuracy of nitrogen nutrition diagnosis among the single wave-band spectral reflectance(R), the logarithm of single wave-band spectral reflectance(LgR), the dual wave-bands spectral reflectance(R1+R2), and the logarithm of dual wave-bands spectral reflectance(LgR1+LgR2) were compared. The results showed that the high negative correlativity between the 6th expanded leaf spectral reflectance and nitrogen content was existed in visible spectrum region. The fittings of the linear regression equation constructed by spectra variables (LgR550+LgR720, Lg(R550+R720)) and nitrogen content respectively were best among them. Sensitive leaves to nutrient profit and loss should be chosen to be regarded as nutrition diagnosis objects of spectral detection in different stages. Spectral sensitive bands to nitrogen nutrition varied with development stages, so more bands of high correlation and more reliable spectral parameters should be selected to construct models with nutrient elements. The results showed that after the logarithm treatment on spectral parameters, whatever either single band or dual bands, the precision of the regression equation was improved and its stability was strengthened. Its observations suggest that it is great potential that using spectral parameters to make nutritional diagnosis for crops, and more suitable spectral parameters or vegetation index for nutritional diagnosis should be investigated and selected. So more research still need to be conducted to test and improve crop nitrogen diagnosis model.