Motor evoked potentials were recorded from the first dorsal interosseous muscle on the dominant side evoked by transcranial magnetic stimulation in 20 healthy volunteers. Amplitude data revealed skewness of distribution and variance heterogeneity. Natural logarithmic transformation of amplitude data results in normal distribution and ensures variance homogeneity. There was a negative first-order autocorrelation between consecutive recordings, which indicates that a stimulus-free interval of 5 s in consecutive stimulations is insufficient to ensure independent events. In a variability study, no systematic changes in latency and amplitude during day-to-day examinations or in consecutive examinations were found. There was a significant change in latency data (p = 0.02) during a gradual increase of constant elements in the stimulation procedure, whereas the amplitude data were unchanged. Estimation of random contributors to the total variability in the magnetic stimulation technique was performed. Intersubject differences were the major contributor to the variability of latency data but not to variability of logarithmic amplitude data. The variance estimates of different examiners' contribution to the variability was 0.02 for latency data and 0 for amplitude data. This indicates that staff members familiar with the principles in magnetic stimulation and following a thorough verbal and written instruction in the standardized examination procedure can perform the stimulation procedure without increasing the data variability.