Soil moisture shows a strong absorption for spectroscopy while soil organic matter and moisture have identical absorption bands. Therefore, the soil moisture causes interference to soil organic matter detection. The study made the following works: firstly, different soil organic matter dynamic spectrums under different moisture content were acquired with visible near-infrared spectroscopy; secondly, different organic matter content features under the same moisture content were analyzed with two-dimensional synchronization correlation spectroscopy. When the soil moisture is 0%, around 600 and 1 660 nm which characterize soil organic matter band appear strong autocorrelation peaks. With moisture content increasing, around 600 and 1 660 nm band disappear and around 1 931, 2 200 and 1 480 nm band appear strong autocorrelation peaks in the near-infrared region. Soil moisture covers information bands which characterize soil organic matter and affects soil organic matter detection; thirdly, the maximum moisture content samples approximately in the filed participated modeling to eliminate the effect of moisture on soil organic matter detection and improve the model prediction accuracy. The anti-moisture interference prediction model which used 550~650 and 1 610~1 710 nm wavelengths by PLS (Partial Least Squares) quantitative analysis method was established to predicate the soil organic matter content under different moisture content. The results are as follows: predicted correlation coefficient, SEP and RMSEP is 0.954, 0.744 and 0.844 respectively. Predicted effect improves significantly. It is that the method can reduce the effect of moisture on soil organic matter detection.