The validity of electromyographic (EMG) data recorded during whole body vibration (WBV) is controversial. Some authors ascribed a major part of the EMG signal to vibration-induced motion artifacts while others have interpreted the EMG signals as muscular activity caused at least partly by stretch reflexes. The aim of this study was to explore the origin of the EMG signal during WBV using several independent approaches. In ten participants, the latencies and spectrograms of stretch reflex responses evoked by passive dorsiflexions in an ankle ergometer were compared to those of the EMG activity of four leg muscles during WBV. Pressure application to the muscles was used to selectively reduce the stretch reflex, thus permitting to distinguish stretch reflexes from other signals. To monitor motion artifacts, dummy electrodes were placed close to the normal electrodes. Strong evidence for stretch reflexes was found: the latencies of the stretch reflex responses evoked by dorsiflexions were almost identical to the supposed stretch reflex responses during vibration (differences of less than 1 ms). Pressure application significantly reduced the amplitude of both the supposed stretch reflexes during vibration (by 61 +/- 17%, p < 0.001) and the stretch reflexes in the ankle ergometer (by 56 +/- 13%, p < 0.01). The dummy electrodes showed almost no activity during WBV (7 +/- 4% of the corresponding muscle's iEMG signal). The frequency analyses revealed no evidence of motion artifacts. The present results support the hypothesis of WBV-induced stretch reflexes. Contribution of motion artifacts to the overall EMG activity seems to be insignificant.