The detection method of the soil nutrients (organic matter and available N, P, K) were analyzed based on the near infrared spectroscopy technology in order to decision-making for precision fertilization. 54 samples with 7m×7m was collected using DGPS receiver positioning in a soybean field. The soil organic matter, available nitrogen (N), available phosphorus (P), available potassium (K) content was determined, the near-infrared diffuse reflectance spectrum of the soil samples were obtained by FieldSpec3 spectrometer. 54 samples were randomly divided into 40 prediction sets and 14 validation sets. After smoothing, the eight principal components of original spectra were extracted by principal component analysis (PCA). Prediction model of soil organic matter, available nitrogen (N), available phosphorus (P), potassium (K) were respectively established with the eight principal component as input and soil nutrients by measured as the output, and the 14 validation samples were predicted. The results showed that the soil organic matter, available nitrogen (N), available phosphorus (P), potassium (K) prediction model were set up with principal component analysis and LS-SVM, which the correlation coefficients between the prediction value and measurement value were 0.8708, 0.7206, 0.8421 and 0.6858, the relative errors of the LS-SVM prediction was smaller and those mean values were 1.09%, 1.06%, 4.08% and 0.69%. The method of soil organic matter content prediction is feasible.