Approximate distributions of clusters of extremes

In a stationary sequence of random variables, high-threshold exceedances may cluster together. Two approximations of such a cluster's distribution are established. These justify and generalize sampling schemes for clusters of extremes already known for Markov chains.