Running is a popular exercise for all age groups. It helps heart and lung functions, enhances muscle strength and control weight. Nevertheless, excessive fatigue and severe injury resulting from inappropriate running poses might reduce the benefits brought by this exercise and stop people from keeping running regularly. In this paper, we design a system that can monitor the running biomechanics, infer running poses, analyze running patterns, and provide both real-time and off-line feedbacks to reduce unnecessary fatigue and unwanted injuries. Common inappropriate running patterns, over-striding, over-pronating and out-sync, are identified and analyzed with the continuous wearable sensor data streams. Two types of correctional feedbacks are designed to provide users appropriate adjustment guidance: vibrating the areas of user body where improper poses are detected and visualizing running video with sensor waveforms to indicate which inappropriate motions trigger the vibration. With reasonable adjustment, running can be a safe and effective activity for a healthy lifestyle.