In this study, a total of 136 paddy soil samples were collected from 17 different fertilization treatments of two short-term field experiments in Jintan and Zhangjiagang, the main grain production region in the middle and lower reaches of the Yangtze River for near infrared (350 ～2 500 nm) -partial least squares (NIR-PLS) regression analysis. Based on the analysis coupled with the cross validation method, a model was established for quantitative analysis of the total carbon, total nitrogen, C/N ratio, available potassium, available phosphorus, electro-conductivity and soil pH obtained by near infrared diffuse reflectance spectroscopy and traditional chemical analysis. R², determination coefficient value, and RSC, ratio of SD (standard deviation of chemical analysis)/RMSECV (root mean square error of cross validation) are two criteria for evaluation of the model. Results show that for the total carbon , total nitrogen , C/N ratio and pH, R²was 0.94, 0.95, 0.97 and 0.92 and RSC was 4.31, 4.35, 5.60 and 3.37, respectively, suggesting that the model is good in prediction. For available potassium, R²was 0.81 and RSC was 2.23, indicating that the model is good, however, for available phosphorus and electro-conductivity, R² was 0.22 and 0.37 and RSC was 0.16 and 1.31, respectively, demonstrating that the model is not so ideal. To sum up, for paddy soil, relevant quality indices can be rapidly predicted through NIR-PLS regression analysis.