A method by which asymmetrical random data can be transformed to equivalent symmetrical or approximately symmetrical data is presented. The symmetrization operation is based on the concept of random thresholding. Information about the probability density or central moments, at least the third-order central moment, is needed to define the constrained optimization problem that specifies the random threshold variable for accomplishing the symmetrization. An example of an application to detection is presented.