The aim of this study was to determine first, if benign fasciculations and those in amyotrophic lateral sclerosis can de distinguished on the basis of their waveforms or firing characteristics, and second to determine how fasciculation parameters evolved with progression of amyotrophic lateral sclerosis. Fasciculation potentials recorded from 63 muscles of 28 patients with definite amyotrophic lateral sclerosis were compared with those from 21 muscles of 11 patients with the benign fasciculation syndrome. In each muscle, at a single site, up to 15 identifiable fasciculation potentials could be recognized. Thus the characteristics of 430 fasciculations from patients with amyotrophic lateral sclerosis and 191 benign fasciculations were analysed. Fasciculation potential amplitude, area, turns, duration, firing interval, indices of waveform variability, evidence of axonal conduction block, evidence of axonal conduction variability and propensity to produce double fasciculations were measured. The waveforms of fasciculations in amyotrophic lateral sclerosis were on average of shorter duration and had a greater number of turns than benign fasciculations, but, although irregular in both conditions, the firing rate in amyotrophic lateral sclerosis was significantly higher. In both conditions, there was evidence of multifocal distal generation of fasciculations, axonal conduction block in the motor unit arborization and of variable axonal conduction. When severe weakness and marked chronic neurogenic change were present on electromyography, the firing rate of fasciculations in amyotrophic lateral sclerosis was higher but fasciculation potential amplitude, area and indices of waveform variability were little changed. Double fasciculations in which the waveforms of the two potentials were the same occurred in both conditions. The intervals were in two bands: an early band with 4-10 ms intervals showed identical waveforms of the two potentials, indicating the region of generation was the same. A second band of double fasciculation occurred in the tibialis anterior at an interval of 30-50 ms. Here, the first fasciculation waveform was variable in shape but the second fasciculation was the same on each occasion, suggesting reactivation of the fasciculation via the F-wave route. Double fasciculations in which the second discharge was different from the first had flat time-interval histograms, indicating no interaction between different fasciculations. In conclusion, benign and malignant fasciculations are not distinguishable on the basis of waveform; highly complex fasciculation potentials can be seen in both conditions. Fasciculation firing rate and the frequency of double fasciculations increases in amyotrophic lateral sclerosis when there is a marked lower motor neuron abnormality.