Abstract: The spectrum technology is one of modern agriculture research focuses, which can be used for realtime, fast, nondestructively monitoring crops nitrogen nutrition condition. The spectral diagnosis models were constructed on the leaf blade level for corn nitrogen nutrition on the base of the whole pooled experimental data. In this study, according to the physiological characteristics of corn (Zea mays L.) nitrogen nutrition, the 6th and 12th (fruit leaf) fullexpanded leaf were acted as observed object using the spectrum technology at the key growth stages. The results show that among 19 spectral characteristic parameters, there are better correlations between leaf nitrogen content and the rededge slope (Dr), green summit maximum reflectance (Rg), ratio vegetation index (RNIR/Red) and normalized difference vegetation index (R(NIR-Red)/(NIR+Red)), respectively. In view of simpleness and practicality of the diagnosis models, ratio vegetation index (RNIR/Red) is chosen to act as spectral variable for leaf N content determination. Thecorrelation relationship between RNIR/Red and different leaf N content is higher in the 6th leaf than that in the 12th leaf at the early growth stage, and higher in the 12th leaf than in the 6th leaf at the late growth stage. Therefore, the 6th fullexpanded leaf is acted as object for nutritional diagnosis at the early growth stage, and the 12th fullexpanded leaf is acted as object for nutritional diagnosis at the late growth stage. Passed through regression analysis and validation, logarithmic model and exponential model with variables RNIR/Red and leaf N content have high reliability and stability at the early growth stage and at the late growth stage, respectively. So those models with variables, RNIR/Red and leaf N content, are regarded as diagnosis models for corn nitrogen nutrition at different growth stages. These results are provided with practical significance for researching and exploiting low cost and portable crop nitrogen nutrition diagnosis instrument.