1. Transcranial cortical stimuli (TCCS) were used to elicit motor responses in contralateral wrist flexor and extensor muscles of healthy adult subjects. The motor responses were assessed by surface EMG recordings, by needle recordings of single motor-unit discharges, and by measurements of wrist twitch force. Our main aim was to analyze the single-unit events underlying those changes in latency, amplitude, and duration of the compound EMG responses, which could be induced by voluntary preactivation of target muscles and by changes in stimulation strength. 2. Different stimulus strengths were tested with and without background contractions in the flexor or extensor muscles. For each test (consisting of a series of 20 stimuli) the compound EMG responses were averaged and displayed together with the averaged wrist force signals. Responses of individual flexor and extensor motor units were displayed in raster diagrams and peristimulus time histograms. For units exhibiting a background firing, the mean background interdischarge interval was calculated and compared with the subsequent poststimulus intervals. 3. In relaxed muscles, a shortening of onset latency of evoked compound EMG responses was observed when raising stimulation strength. A similar latency reduction was not seen in any of the single-unit recordings. This would be consistent with the size principle of motoneuron recruitment. 4. A shortening of onset latency of evoked EMG potentials was observed also as a result of a voluntary preactivation. Such latency shifts, which were seen also in single-unit recordings, might be attributed to variations in the time required for D and I wave temporal summation at the anterior horn cell. 5. When raising stimulation strength or when adding voluntary background contraction, the evoked compound EMG potential grew not only in amplitude but also in duration, as later peaks of activity were added to the initial ones. Under optimal conditions (strong stimulus + background contraction), the period of excitation (termed E1) had an onset latency of approximately 15 ms and a duration of approximately 35 ms and was similar for wrist flexor and extensor muscles. 6. We never saw the same flexor or extensor unit fire more than once during the E1 period. For units preactivated by a background contraction, the stimulus-triggered impulse exhibited latency shifts, which, to a large extent, depended on the timing of the stimulus in relation to a preceding background discharge and which could be influenced by a change in stimulation strength.(ABSTRACT TRUNCATED AT 400 WORDS)