Visible/near-infrared spectroscopy (Vis/NIRS) was investigated for determination of soil properties, namely, available nitrogen (N) and available potassium (K). In order to improve the predictive precision and eliminate the influence of uninformative variables for model robustness, Monte Carlo uninformative variables elimination (MC—UVE) methods were proposed for variable selection in available N and available K NIR spectral modeling. Partial least squares (PLS) models analysis was implemented for calibration models. The modeling variable number was reduced to 210 from 751 for available N calibration model and 150 for available K calibration model. The performance of the model was evaluated by the correlation coefficient (R), RMSEP. The optimal MC—UVE PLS models were achieved, and R, RMSEP were 0.84, 17.1mg/kg for N and 0.76, 15.4mg/kg for K, respectively.