The effect of soil moisture on the determination of red soil organic matter (SOM) is a theoretical basis for the rapid predication of SOM in the flied. Spectral measurements (350-2 500 nm) of soil samples under different moistures were conducted in a controlled laboratory environment and partial least-squares regression (PLSR) was used to establish SOM prediction model. The results showed that the correlation of SOM and first order derivative reflectance firstly were increased, and then decreased as soil moisture increased, the highest correlation occurred when soil moisture was at 100-150 g/kg. Spectral regions of 380-2 400 nm, 380-1 300 nm and 1 300-2 400 nm were selected to establish SOM prediction models under different soil moistures. The accuracy of prediction model were increased firstly and then decreased with the increase of soil moisture. Using 1 300-2 400 nm to build SOM prediction model could avoid the influence of iron oxide and get highest prediction accuracy. The results demonstrated that when soil moisture was less than 200 g/kg, SOM of red soil could be predicted by PLSR using first order derivative reflectance in 1 300-2 400 nm under a controlled environment.