This work presents the use of unscented transforms(UT) for the description of statistical uncertainty inelectromagnetic coupling between cables. UT greatlyreduce the computational burden for the statistical analysisof nonlinear problems compared with more traditionalapproaches such as the Monte Carlo technique. Couplingbetween cables has a nonlinear parameter dependence andhas a high variability due to the variability in the cablebraid manufacture and the highly variable nature of cablelayout. Therefore, cable coupling can only be definedwithin statistical limits. First, it is shown that by analyzing the resonances the important features of maximum coupling and the point of maximum coupling can be characterized. It is then demonstrated how UT can be used to efficiently identify the parameters which contribute significantly to the uncertainty in cable coupling and then to provide a measure of the probability distribution for the multivariate problem.