The runoff loss of nitrogen and phosphorus in farmland results in severe surface water pollution. Runoff losses of nitrogen and phosphorus could be effectively assessed with the empirical model. Using the power function model with the LTS and S estimation, we fitted and evaluated the regression relationship between runoff and runoff losses of soil erosion and various nitrogen and phosphorus under different crop covers at Xinmaqiao Experimental Station located in Huaibei Plain of China. The results showed a good power function of regression between runoff and losses of soil erosion and nitrogen, respectively, and the coefficient of determination r(2) varied from 0.52-0.94 and followed the order of soil erosion > total nitrogen loss > dissolved nitrogen loss > particulate nitrogen loss. The fitting exponent parameter b of the soil erosion in fallow, cotton, and corn fields and nitrogen loss in fallow field were all greater than 1.0, and the soil erosion and nitrogen loss from these fields could be controlled by reducing the runoff. High phosphorus concentrations in corn, cotton, and soybean fields occur during small runoff events, and concentration of particulate phosphorus does not change much with increased runoff, whereas the concentrations of total phosphorus and dissolved phosphorus decrease. The concentrations of total phosphorus, dissolved phosphorus, and particulate phosphorus in fallow fields increase with runoff.