Abstract-- This paper considers uncertain constrained systems, and develops a method for computing a probabilistic output admissible (POA) set which is a set of initial states probabilistically assured to satisfy the constraint. We first present the basic algorithm for computing the POA set. Though the algorithm is easily obtained by utilizing the sequential randomized algorithm, its computational effort is enormous and the termination in finite iterations is not assured. Thus, we develop an improved algorithm whose computational effort is reasonably small and the finite termination is guaranteed. Numerical simulations demonstrate the effectiveness of the present method. Keywords-- Constrained system, Maximal output admissible set, Uncertainty, Randomized algorithm