Roughness index is one of the most relevant pavement performance indices due to its impact on driving safety and comfortability. This study applies the newly innovated response-based device (RBD) and the corresponding adjusted accelearation root-mean-square index (AARI) to develop the roughness prediction model for National Freeway system. Roughness measurements of a total length of 58.8 km for both directions of northern section of National Sun Yat-Sen Freeway were conducted periodically during study period and the traffic and weight information were also collected. It is concluded that not only the RBD can provide economic and efficient roughness data collection, but the developed prediction model serves as the pre-screening function for further pavement inspection and maintenance needs.