Methods for analyzing various motion artifacts in photoplethysmography (PPG) signals, recorded by a wristworn device are reported. The analysis looks both at intrinsic PPG signal properties, through standard deviation, skew and kurtosis, but also at its relationship to five possible motion reference signals, through the correlation coefficient and mutual information. The investigated references include the X, Y, Z accelerometer axes, acceleration magnitude and a different wavelength PPG channel. The latter showed higher correlation to the input PPG signal during movement. A wavelet based motion artifact reduction algorithm is described. The algorithm improves the visibility of the heart rate (HR) components in the frequency domain representation of the signals.