This article presents a method to characterize stochastic observables defined by induced surface currents and fields in electromagnetic interactions with uncertain configurations. As the covariance operators of the stochastic distributions and fields are not compact, a strict Karhunen-Loeve (KL) approach is not possible. Instead, we apply a point-spectrum regularization by expanding the stochastic quantities on a finite-element-like basis. The coefficients of the KL expansion are approximated analytically in a polynomial-chaos (PC) expansion. The novelty of our approach resides in its ability to handle multiple PC expansions simultaneously and determine the orders of the KL and PC expansions adaptively. Thismethod is illustrated through the example of the voltage induced at the port of a random thin-wire frame illuminated by random plane waves. The results show the accuracy and computational efficiency of the proposed method, which provides a complete characterization of the randomness of the observable.