If topographical EEG is to be a useful tool for localising cerebral processes, then the results of the same, or closely similar experiments, using different samples should yield similar results. Although the reliability of EEG is well established in other ways, there is little available data on the reproducibility of EEG topography across experiments. The aim of this study was to determine the reproducibility of topographical EEG by comparing the results of two independently conducted experiments. EEG was recorded during an Eyes Open baseline and a motor task condition (the Luria finger opposition task) in two independent samples of healthy subjects. EEG was recorded in 2.56-s epochs and analysed by FFT into conventional theta, alpha and beta 1 frequency bands. The EEG amplitude for each subject in each frequency range was averaged over a minimum of 60 s. Separate group averages for each sample were calculated and the resulting topographical distributions of electrical potential and current density were compared. The results indicated that the reproducibility of electrical potential in the theta and beta 1 frequency ranges was extremely poor and only approached acceptable levels in alpha. Reproducibility of current source density was poor in all frequency ranges. Although some improvement in reproducibility was obtained following spatial smoothing for alpha potential, the highest reproducibility achieved was only 0.65. Reasons for the poor reproducibility of topographical EEG and the implications of these findings are discussed.