Abstract: ECG was an important basis for clinical diagnosis of diseases.But the noises,such as baseline wander,affected the accuracy of ECG diagnosis.According to the properties of baseline wander and intrinsic mode function (IMF),a new method called adaptive filtering based on empirical mode decomposition (EMD) combined morphological filtering was proposed in the paper.First,ECG signal was decomposed into a series of IMFs by EMD,then those IMFs selected were processed further by the morphological filter.The filtering result was regarded as the reference input signal.Finally,the error output signal was ECG signal removed the baseline wander.The proposed method was compared with common EMD,adaptive filter based on EMD,and it was tested by using MIT-BIH Database.The experiment results showed that the method was effective in removing ECG baseline wander.