Abstract: This paper sets up a supply chain performance prediction model based on rough sets and Support Vector Regression machine（SVR） from knowledge discovery and data mining perspective.According to a supply chain performance prediction example，firstly its index based on the balanced scorecard system is reduced，and then the reduction index is input to SVR for training.Finally，the forecast sample is input to the model and supply chain performance predictive value is got.The forecast result is matched with the actual result.