Background: Response Elaboration Training (RET; Kearns, 1985) has been found to consistently result in increased production of content in discourse with persons with aphasia. Positive treatment effects have been reported for persons representing a variety of aphasia types and severities. RET was modified for application with persons with acquired apraxia of speech and aphasia and positive outcomes were also associated with the modified treatment (Wambaugh & Martinez, 2000). Although RET has received systematic study, its stimulus generalisation effects are not well understood. Aims: This investigation was designed to measure the stimulus generalisation effects of modified RET (M-RET) in a variety of conditions as well as to further study the effects of M-RET applied to a personal recount condition. Methods & Procedures: Multiple baseline designs (across behaviours and participants) were utilised to examine treatment effects. Treatment was applied sequentially to picture sets and a personal recount condition with six persons with chronic aphasia. Production of correct information units (CIUs) was measured in the following conditions: (1) discourse production in response to sets of trained and untrained pictures, (2) home conversations, and (3) production of discourse in structured tasks. Formal measures of functional communication were also completed prior to and following treatment. Outcomes & Results: Increases in production of CIUs in response to pictures were observed for 11 of the 12 applications of M-RET to picture sets. Response generalisation to untrained picture sets was associated with M-RET applied to pictures sets; increases were slight and were greater for untrained sets that were probed more frequently. Maintenance of gains was generally strong for the participants with nonfluent aphasia, but was minimal for the participant with fluent aphasia. Gains were not evident for M-RET applied to personal recounts; only one participant evidenced changes possibly associated with treatment in the personal recount condition. Improvements in structured discourse samples and a functional communication measure were observed for the majority of the participants following treatment. Lack of compliance in completion of recordings of home conversations limited the utility of that measure. Conclusions: M-RET applied to pictures resulted in improvements in production of content in treated and untreated picture conditions for the majority of the participants. Treatment effects extended to additional outcome measures. Although some positive changes were observed for the participant with fluent aphasia, maintenance was problematic. Application of M-RET to a personal recount condition was not associated with improved performance for most of the participants.