Dynamic balance control during human walking can be described by the distance between the mediolateral (ML) extrapolated center of mass (XCoM) position and the base of support, the margin of stability (MoS). The ML center of mass (CoM) position during treadmill walking can be estimated based on kinematic data (marker-based method) and a combination of ground reaction forces and center of pressure positions (GRF-based method). Here, we compare a GRF-based method with a full-body marker-based method for estimating the ML CoM, ML XCoM and ML MoS. Fifteen healthy adults walked on a dual-belt treadmill at comfortable walking speed for three minutes. Kinetic and kinematic data were collected and analyzed using a GRF-based and marker-based method to compare the ML CoM, ML XCoM and ML MoS. High correlation coefficients (r > 0.98) and small differences (Root Mean Square Difference < 0.0072 m) in ML CoM and ML XCoM were found between the GRF-based and marker-based methods. The GRF-based method resulted in larger ML XCoM excursion (0.0118 ± 0.0074 m) and smaller ML MoS values (0.0062 ± 0.0028 m) than the marker-based method, but these differences were consistent across participants. In conclusion, the GRF-based method is a valid method to determine the ML CoM, XCoM and MoS. One should be aware of higher ML XCoM and smaller ML MoS values in the GRF-based method when comparing absolute values between studies. The GRF-based method strongly reduces measurement times and can be used to provide real-time CoM-CoP feedback during treadmill gait training.