We wished to develop a noninvasive electrophysiological measure of functioning of human cerebral cortex. The response chosen was that elicited by a corticocortical afferent input, specifically the transcallosal response (TCR). We used ourselves as subjects. The large shock artifacts associated with passing large currents through the head were reduced by: stimulating percutaneously with a focal anode (5 cm2) situated midway between F4 and C4 (i.e., frontal region, standard 10-20 nomenclature) and a medially located, semicircular grounded cathode (26 cm2) no closer than 3 to 4 cm away; and recording with the focal scalp electrode over homologous left frontal cortex vs. two or more reference electrodes interconnected through variable resistors and at least 6 cm away. Transcranial stimulation with pulses of 80 to 140 mA and lasting 100 microseconds elicited a brief diphasic deflection (latency, 2 to 4 ms), followed by an initially positive, often bifid wave with a latency of 9 to 14 ms, lasting 18 to 44 ms and reaching 5 to 10 microV. The initial diphasic deflection greatly increased when the stimulating anode was moved laterally over temporalis muscle, indicating an electromyographic (EMG) origin. The EMG, but not the positive wave, was elicited when the interpolar distance was reduced to 1 cm, which greatly reduced the fraction of current traversing the cortex (epicranial stimulation); furthermore, occluding the circulation to the scalp for 30 minutes by rubber tubing above the zygomas and brows dulled sensation, reduced the EMG, but did not alter the latency or rising phase of the positive wave. Early somatosensory components, if present, were small compared with the positive wave.(ABSTRACT TRUNCATED AT 250 WORDS)