This study evaluated the extent to which measures of lexical diversity (type-token ratio and number of different words produced) differentiated children with specific language impairment (SLI) from children whose language skills were following typical developmental expectations. Analysis of 50- and 100-utterance samples revealed that children with SLI did not significantly differ from their age- and language-equivalent peers on type-token ratio; however, children with SLI did use significantly fewer different words than their age-equivalent peers in these samples. When samples of 100 and 200 tokens were considered, the children with SLI also used significantly fewer different words than their age-equivalent counterparts. Overall, the findings of this study discourage the use of type-token ratio, as traditionally calculated, in many clinical and research activities related to children with SLI. In contrast, the number of different words produced provides a more sensitive and informative estimate of lexical diversity.