Motion artifacts (MA) have long been a problem in biopotential measurements. Adaptive filtering is widely used for optimal noise removal in many biomedical applications. However, the existing adaptive filtering methods involve the use of additional sensors, limiting the applicability of adaptive filtering for MA reduction. In the present study, a novel adaptive filtering method without need for additional sensors is proposed. In biopotential measurements, movement of the electrodes and their leads may cause variations not only in the skin and half-cell potential (motion artifacts), but also in the electrode-skin impedance. Such impedance variations may also cause power-line interference modulation (PLIM), resulting in additional spectral components around the power-line interference (PLI) in the frequency domain. Demodulation of the PLI may reflect the movement-induced electrode-skin impedance variation, and can therefore represent a reference signal for the adaptive filter. Preliminary validation on ECG measurements with seven volunteers showed a high correlation coefficient (R = 0.97) between MA and PLIM, and excellent MA removal by the proposed adaptive filter, possibly leading to improved analysis of biopotential signals.