Purpose: To investigate the usefulness of 3 tasks known to be effective diagnostic clinical markers of specific language impairment (SLI) in children: (a) nonword repetition, (b) sentence repetition, and (c) grammaticality judgments of finiteness marking. Method: Two groups of young adults, 13 with SLI and 18 with typical language, completed 3 experimental tasks: (a) nonword repetition, (b) sentence repetition, and (c) grammaticality judgments of sentences that were either correct or contained an omitted finiteness marker, an overt agreement error, or an omitted progressive -"ing." Analyses included receiver operating characteristic curve analyses and computation of likelihood ratios associated with the use of each task as a clinical marker for SLI, as well as development of a logistic regression model that used multiple tasks as predictors. Results: Each marker task significantly contributed to classification of adults as affected or unaffected by SLI, with moderate positive and negative likelihood ratios. A combination of the 3 marker tasks was the best predictor of affectedness status with moderate to large likelihood ratios. Conclusions: The results suggest that SLI persists into adulthood and that effective clinical markers of this disorder are similar to those used to identify SLI in children. Refinement of these tasks to increase their likelihood ratios will improve their usefulness in diagnosing SLI in adults.