Motion artifact is the main interference of dynamic ECG monitoring and will bring misdiagnosis or miss-diagnosis. In order to solve this problem, an accelerometer aided dynamic ECG monitoring system is designed. This system collects Lead V2 ECG signal and abdominal acceleration. Patient’s daily activity is estimated from his/her acceleration data and motion artifact contaminated ECG is labeled, then use PQRST feature points detection and RLS adaptive filter algorithm to remove the motion artifact. The effectiveness of this algorithm is verified by actual ECG signals, daily activities such as coughing, squatting and sitting can be recognized and motion artifact caused by such activities can be removed.