Purpose: The authors aimed to establish whether 2 theoretically motivated interventions could improve use of verb argument structure in pupils with persistent specific language impairment (SLI). Method: Twenty-seven pupils with SLI (ages 11;0-16;1) participated in this randomized controlled trial with "blind" assessment. Participants were randomly assigned to 1 of 3 therapy groups: syntactic-semantic, semantic, and control. All pupils received 9 weekly half-hour individual therapy sessions. They were assessed on a specifically designed video test pretherapy, posttherapy, and at follow-up. Results: Pupils receiving the syntactic-semantic and semantic therapies made significant progress (d greater than 1.0), which was maintained at follow-up and generalized to control verbs. Both therapies improved linking of arguments to syntax, and the syntactic-semantic therapy tended to increase use of optional arguments. Pupils receiving the control therapy made no progress. Conclusion: Both methods of argument structure therapy were effective. Comparisons of their effectiveness in specific areas led to the hypotheses that the pupils' initial difficulties with linking resulted from ill-defined semantic representations, whereas their limited use of arguments may have resulted from syntactic difficulties. When therapy is theoretically grounded, it can inform theories, be time limited, and be effective for older children with SLI. The following are appended: (1) Scenes in video test; and (2) Shapes, colors, questions, and symbols for syntactic-semantic therapy. (Contains 4 tables and 7 figures.)