A new ST segment analysis scheme is developed for helping the doctors to browse the electrocardiogram (ECG) signals rapidly and then give a correct diagnosis. The preprocessing consists of baseline wander attenuation using median filter, high frequency noise rejection using order statistic filter, and then QRS complexes detection using wavelet analysis. In this paper, after the beginnings and the ends of the ST segments are mutually chosen by doctor, we compute the total deviations between ST segments and the modified baselines using mean value, and then plot the whole trend of the deviations. Doctors can browse the original ECG signals handily by our browsing system, and then give a diagnosis colligating other clinic features of the patient.