We investigate the problem of enumerating the words that are common to two random strings. We show that when the source models are memoriless that the number of common words is sublinear in the length of the sequence, and linear when the source models are exactly the same. We draw the same conclusions for the number of common nodes in the associated respective suffix trees.