Recent findings suggested that the choice of the EEG reference might be a critical issue for the study of anterior asymmetry in the alpha band. The present paper investigates the validity of different reference schemes for the measurement of alpha asymmetry. A 32-channel resting EEG was recorded with a common vertex reference (Cz), and transformed into computer-averaged ears (A1 + A2), average reference (AR), and current source density derivations. A correlation analysis of an alpha asymmetry measure between all derivation schemes indicated a poor convergent validity for anterior sites but an excellent convergent validity for posterior sites. Further analyses suggested the presence of substantial alpha activity at the various reference sites (Cz, A1 + A2, AR), which might be similar in magnitude to anterior but smaller than posterior alpha. These findings suggest that the validity of a reference scheme is a function of the signal-to-noise ratio of the electrical activities at target and reference sites. The limitations of each reference scheme for the measurement of anterior alpha asymmetry are discussed.