The site of the recording electrode influences the amplitude of the compound muscle action potential (CMAP) and its variation over a length of nerve. The effects of large electrodes on this source of intraindividual variability were assessed. Right median nerves of 20 healthy subjects were studied, and recordings made at three sites (at 1-cm intervals) using five electrode sizes (0.01, 1, 2, 4, and 10 cm2). Site-induced variability was defined as the standard deviation (SDi) and coefficient of variation (CVi) of the measurements of the three sites. Site induced variability of all parameters (latency, duration, amplitude, area, MNCV, and the percentile changes of duration, amplitude, and area over the forearm) decreased significantly with electrode size. Decreases were most pronounced for amplitude and area: CVi fell from 29% and 30% (0.01-cm2 electrode) to 10% and 8% (10 cm2). It is argued that large electrodes record activity of more motor units than small electrodes, and that their measurement fields overlap to a greater extent. The use of large electrodes is recommended in order to reduce site-induced CMAP variability.