This report addresses some of the statistical problems that are encountered when the test/retest recording reliability of fatigue-related parameters of the EMG power spectrum is evaluated. It can be shown that some classical methods for reliability assessment such as correlational procedures are unsuitable for this purpose. Because the EMG power spectrum fatigue parameter depends on metabolic changes in muscle tissue, it is suggested that methods similar to those used in the evaluation of bioequivalence studies may be more appropriate in the assessment of such test/retest results.