Efficient business operation must monitor the performance of a firm and make effective forecasting. This study builds a forecasting model for the performance of company X which is one of the leading TFT-LCD companies in Taiwan. Major factors affecting its revenue are identified. The proposed forecasting model is developed by applying a set of quantitative methods. Time series analyses such as moving average and smoothing techniques are employed. In addition, three predicators are applied in a regression model and a back-propagation neural network model respectively. Mean absolute percentage error is used to evaluate and select the best forecasting model accordingly. Experiments show that the proposed forecasting model performs well in practice.