Distraction driving has been reported to significantly increase accident risks. In fact, taking eyes off from road is defined as distraction driving by United States Department of Transportation (USDOT) and usually comes with abnormal head poses. This paper presents a distraction driving monitoring system based on dual compass head motion sensoring technology. The major difficulty is that single headmounted sensoring device suffers from vehicle motion. When vehicle is moving, the sensory angle velocity from headmounted compass includes vehicle motion, but not pure driver's head motion. The affected head angular velocity causes inaccurate subsequent analysis. Therefore, an additional compass was placed on vehicle for calibration and compensation. Then, abnormal head poses were applied to calculate distraction index. When distraction index was greater than a threshold, driver was warned to make sure driving safety. In addition, the distraction events can be feedback to driving behavior analysis platform. According to the collected data, the abnormal head pose events not only can be real-time warned to driver, but also can mine the common distraction patterns based on collaborative data analysis techniques. These data can be notified to driver for early reminder when drivers are approaching the common distraction road sections.