A critical clinical issue is the identification of a clinical marker, a linguistic form or principle that can be shown to be characteristic of children with Specific Language Impairment (SLI). In this paper we evaluate, as candidate clinical markers, a set of morphemes that mark Tense. In English, this includes -s third person singular, -ed regular past, BE, and DO. According to the Extended Optional Infinitive Account (EOI) of Rice, Wexler, and Cleave (1995), this set of morphemes is likely to appear optionally in the grammars of children with SLI at a rate lower than the optionality evident in younger controls. Three groups of preschool children participated: 37 children with SLI, and two control groups, one of 40 MLU-equivalent children and another of 45 age-equivalent children. Three kinds of evidence support the conclusion that a set of morphemes that marks Tense can be considered a clinical marker: (a) low levels of accuracy for the target morphemes for the SLI group relative to either of the two control groups; (b) affectedness for the set of morphemes defined by the linguistic function of Tense, but not for morphemes unrelated to Tense; and (c) a bimodal distribution for Tense-marking morphemes relative to age peers, in which the typical children are at essentially adult levels of the grammar, whereas children in the SLI group were at low (i.e., non-adultlike) levels of performance. The clinical symptoms are evident in omissions of surface forms. Errors of subject-verb agreement and syntactic misuses are rare, showing that, as predicted, children in an EOI stage who are likely to mark Tense optionally at the same time know a great deal about the grammatical properties of finiteness and agreement in the adult grammar. The findings are discussed in terms of alternative accounts of the grammatical limitations of children with SLI and implications for clinical identification.