In 23 healthy adult volunteers motor action potentials (MAPs) were elicited in upper and lower limb muscles during stimulation of appropriate sites at spinal and scalp level, through skin electrodes. 'Bifocal' stimulation of scalp and spine motor tracts was performed with 2 plaques (3.5 cm2 each), delivering single pulses of 440-940 mA, less than 50 microseconds in duration, which elicited high voltage (up to 10 mV) MAPs in arm and leg muscles. 'Unifocal' stimulation of scalp was carried out through a cathode consisting in a belt or in a series of rectangular interconnected plaques secured around the head, 1-2 cm rostral to the nasion-inion plane, and in a circular anode placed on the appropriate scalp site. MAPs with similar amplitude-latency characteristics were recorded with both 'bifocal' and 'unifocal' stimulating methods. However, the 'unifocal' stimulation necessitated 5-10 times less current than the 'bifocal' one. The 'unifocal' device using the interconnected plaques (6-12 in number) provided the most tolerable stimuli with the lowest amount of current (60-106 mA, rectangular pulses of 100-150 microseconds). Conduction times and velocities of motor pathways in various 'central' and 'peripheral' districts were calculated. Voluntary contraction of target muscles remarkably enhanced MAP amplitudes during scalp, but not during spine stimulation. A nerve action potential was recorded from ulnar nerve during scalp stimulation. MAPs in hand muscles to scalp stimulation were obliterated by the simultaneous activation of the peripheral fibres innervating the target muscle, because of collision between ortho- and antidromically propagated motor impulses. Anodal stimuli showed liminal values significantly lower than the cathodal ones. Mapping studies have been carried out with 'unifocal' scalp stimulation by using different types of anode and of stimulus parameters.