In this study, the LMS algorithm was employed to detect and track baseline wanders introduced during the acquisition of electrocardiogram signals. The baseline wander was removed from the signal and ECG is corrected. For this purpose, the baseline signal was modeled as a constant and a line in the LMS algorithm. The two approaches were applied to the signals received from MIH-BIH database records and the results were reported and compared. When the experimental results are investigated it is observed that the line model adapts faster and tracks baseline wander.