Variability of the amplitude of motor evoked potentials was studied in 33 multiple sclerosis (MS) patients and 21 healthy subjects. Normal probability plots revealed skewness of the distribution of amplitudes obtained from the abductor pollicis brevis (APB) muscle during muscle relaxation, muscle contraction, and by paired magnetic stimulation. Natural logarithmic transformation of amplitude data resulted in normal distribution. Negative first-order autocorrelations were established for consecutive recordings independent of the repetition rate tested (5 s, 8 s, 12 s, and 20 s). A given measurement had a tendency to be systematically followed by a measurement of a lower value, indicating that magnetic stimulations induce a long-lasting inhibitory effect on the excitability of the motor pathways. A significant decrease in amplitude variability was achieved by a controlled muscle contraction of the target muscle compared with an uncontrolled muscle contraction in MS patients. Paired magnetic stimulation halved the coefficient of variation of amplitude data compared with amplitude obtained during muscle relaxation, and is introduced to reduce amplitude variability in conditions where no collaboration is possible.