One of the most important aspects of the electrocardiogram (ECG) signal processing is the removal of noises from the signals. In the present work a filter bank architecture based on wavelet transform is proposed for this purpose. Proposed design uses four levels of wavelet transform based filter bank for the ECG signals denoising. A digitized ECG signal is applied as an input to the four levels of wavelet transform based filter bank that separates the ECG signal from the noises. Obtained results show that the proposed architecture requires less hardware, small area on the chip and lower cost compared to the previously designed architectures. © 2017 IEEE.