The coefficient of variation or relative standard deviation is a measure of a device’s noise–to–signal ratio. It is an increasingly popular metric used by manufacturers and requested by review agencies to determine whether two distinct systems exhibit comparable variability. In this report, an exact method is developed to test the statistical equality of the coefficients of variation for two different devices. Moreover, the method is extended to test for statistical equivalence within a clinical margin. A discussion of the technical subtleties required in the calculation of CV variance is provided. Examples of simulated and clinical data are examined.