Neuromuscular electrical stimulation (NMES) involves artificially activating the muscle with a protocol designed to minimize the discomfort associated with the stimulation. Typical settings of NMES exercise involve the application of electrical stimuli delivered in intermittent trains through surface electrodes positioned over the muscle motor point and pre-programmed stimulation units. Since the 18th century this method has been used either to supplement or to substitute for voluntary activation of muscle in many rehabilitation settings, e.g., for muscle strengthening, maintenance of muscle mass and strength during prolonged periods of immobilization. Despite the lack of scientific evidence regarding the physiological adaptations to NMES exercise – particularly for impaired muscles – recent studies have demonstrated the effectiveness of NMES programs for the improvement of both physical and muscle function in orthopaedic, cardiovascular, neurology, and musculoskeletal patients. The application of NMES results in metabolic and neuromuscular effects – both acute and chronic – which are extremely different compared to voluntary muscle activation. A good knowledge of the characteristics of the electrical current, as well as of the specific physiology of electrically evoked contractions, is required to optimize the effectiveness of NMES treatments and to minimize the risks associated to this procedure. Therefore, after a quick overview of the methodological basis of NMES, I will focus on two aspects of NMES physiology that have been the subject of numerous discussions in the last 20 years: the involvement of the nervous system during peripheral NMES and the differences in motor unit recruitment pattern between NMES and voluntary contractions. Surface NMES is generally considered a technique to activate muscles without involving the nervous system. However, the magnitude of central effects evoked by the peripheral stimulation can be substantial. First, NMES evokes a muscle contraction by activating intramuscular branches of the nerve to the muscle and not the muscle fibres directly. Second, in intact systems NMES does not actually bypass the peripheral nervous system – because of the bilateral propagation of action potentials along the stimulated axons – and would even activate spinal neurons via reflex inputs to the spinal cord (see Collins et al. 2007) and selected brain regions in a dose-response manner (Smith et al. 2003). Third, a series of fatigue and training studies completed in our laboratory have demonstrated that NMES could evoke widespread activity within the central nervous system that is capable of mediating a range of acute and chronic neural adaptations, e.g., increased muscle activation (Maffiuletti et al. 2002), cross-education effect. When skeletal muscles are artificially activated, as with NMES, the involvement of motor units is quite different from that underlying voluntary activation. The main argument in favour of such a difference is that large-diameter axons are more easily excited by electrical stimuli, which would alter the activation order compared with voluntary contractions. However, human experiments yielded contradictory findings with some studies suggesting preferential/selective activation of fast motor units with NMES, and others demonstrating minimal or no difference between the two contraction modalities. We and others (Gregory & Bickel, 2005) have recently suggested that NMES-induced motor unit recruitment is synchronous, spatially fixed (rather superficial), and nonselective/random, i.e., muscle fibres are recruited without obvious sequencing related to fibre types, and therefore NMES can be used to activate fast (in addition to slow) motor units at relatively low force levels. The main consequence of such a particular motor unit recruitment pattern is the exaggerated metabolic stress of a stimulated contraction compared to a voluntary action of the same intensity (Vanderthommen et al. 2003), which inevitably results in greater (and earlier) muscle fatigue. The above-discussed physiological differences between electrically evoked and voluntary contractions constitute an argument in favour of the combination of these two modalities of activation in the context of rehabilitation.