Inhibitory reflexes during voluntary contractions are well described; however, few studies have attempted to use such reflex-mechanisms to modulate electromyographic (EMG) activity in jaw-closing muscles during sleep. The aim was to apply a new intelligent biofeedback device (Grindcare(R)) using electrical pulses to inhibit EMG activity in the temporalis muscle during sleep. Fourteen volunteers participated who were aware of jaw-clenching activity as indicated by complaints from sleep partner, soreness or pain in the jaw-muscle upon awakening and tooth wear facets. The EMG activity was recorded from the temporalis muscle, online analysed and the frequency content determined using a signal recognition algorithm. Based on specific individual parameters for pattern recognition, an electrical square-wave pulse train, which was adjusted to a clear, but non-painful intensity (range 1-7 mA) was applied through the EMG electrodes, if jaw-clenching activity was detected. All volunteers had baseline EMG recordings for five to seven consecutive nights, followed by 3-weeks EMG recordings with the feedback turned on, 2 weeks without the feedback and finally 3 weeks with the biofeedback on. There were no session effects on the average duration of sleep hours (P = 0.626). The number of EMG episodes/hour sleep was significantly reduced during the two sessions with biofeedback (54 +/- 14%; 55 +/- 17%, P < 0.001) compared with baseline EMG activity and the session without biofeedback. The present study suggests that biofeedback with electrical pulses does not cause major disruption in sleep and is associated with pronounced reduction in temporalis EMG activity during sleep.