1. Measurements of twitch contractile properties of human motor units recorded by spike-triggered averaging may be distorted by partial fusion between twitches, because motor units seldom fire at rates below 8-10 Hz. The effects of this fusion were examined by comparing the responses of 27 human thenar motor units when their motor axons were stimulated at 1, 8, and 10 Hz. 2. Resultant forces were calculated from the abduction and flexion force components, together with various contraction and relaxation rate indexes as reported previously. Values for single twitches were compared to measurements made from the unfused force fluctuations ("apparent twitches") of the same units recorded during 8 and 10 Hz stimulation. 3. For all units, stimulation at 8 and 10 Hz caused partial twitch fusion. At 10 Hz, mean values for "apparent twitch" forces, contraction times (CT), and one-half relaxation times (1/2RT) were reduced to 44, 76, and 52% of the corresponding values measured from separate twitches evoked by 1 Hz stimulation. Similar but smaller reductions were seen at 8 Hz. 4. Slow units, with initial twitch CT greater than 60 ms, showed significantly more distortion of all "apparent twitch" parameters when stimulated at both 8 and 10 Hz, compared to fast units (less than 50 ms). 5. The potentiated abduction force component data were compared with abduction forces obtained previously by spike-triggered averaging from the same muscle group. Mean force obtained by spike-triggered averaging ("STA twitch" force: 21 mN) was significantly larger than that measured in abduction in response to either 1 or 10 Hz motor-axon stimulation (14 mN, 6 mN, respectively).(ABSTRACT TRUNCATED AT 250 WORDS)