Contour ridge systems may lead to seepage that could result in serious soil erosion. Modeling soil erosion under seepage conditions in a contour ridge system has been overlooked in most current soil erosion models. To address the importance of seepage in soil erosion modeling, a total of 23 treatments with 3 factors, row grade, field slope and ridge height, in 5 gradients were arranged in an orthogonal rotatable central composite design. The second-order polynomial regression model for predicting the sediment yield was improved by using the measured or predicted seepage discharge as an input factor, which increased the coefficient of determination (R2) from 0.743 to 0.915 or 0.893. The improved regression models combined with the measured seepage discharge had a lower P (0.007) compared to those combined with the predicted seepage discharge (P=0.016). With the measured seepage discharge incorporated, some significant (P<0.050) effects and interactions of influential factors on sediment yield were detected, including the row grade and its interactions with the field slope, ridge height and seepage discharge, the quadratic terms of the field slope and its interactions with the row grade and seepage discharge. In the regression model with the predicted seepage discharge as an influencing factor, only the interaction between row grade and seepage discharge significantly affected the sediment yield. The regression model incorporated with predicted seepage discharge may be expressed simply and can be used effectively when measured seepage discharge data are not available.