Surface electromyogram usually incurs severe influence of strong electrocardiography (ECG) signal. This paper presents a novel denoise method using independent component analysis (ICA) to remove the interference of ECG from sEMG recordings. The present study removed the ECG in independent components decomposed from the sEMG by a high-pass filter that kept the useful information as more as possible during ECG cancellation. The filtering ICA ECG cancellation results were compared with the simulated pure sEMG in time and frequency domain, which suggested that the ICA ECG cancellation with 30Hz high-pass filter would be the most appropriate method to extract the useful sEMG signals from multi-channel sEMG measurement. The kurtosis is used to measure the nongaussianity of signal; it may beused as a standand for automatic determination of the independents of ECG or EMG.