In the worldwide WLAN device maker and design’s area, Taiwan plays an important role in the market. The development of WLAN standard is rapid and the market is competitive. For all WLAN device companies, how to shorten production time to increase efficiency and to reduce cost is an important target. In this study, according to the test data of the production process at one WLAN device maker, we apply Back-Propagation Neural Networks to forecast WLAN transmission efficiency. It is expected to eliminate the test procedure to reduce the production time and cost. In this study, we used the Coefficient of Correlation to prune the structure of the trained network and obtained an effective forecasting result.