In this paper we study a discrete version of the classical team decision problem. It has been shown previously that the general discrete team decision problem is NP-hard. Here we present an efficient approximation algorithm for this problem. For the maximization version of this problem with nonnegative rewards, this algorithm computes decision rules which are guaranteed to be within a fixed bound of optimal.