Background: Capturing evidence of the effects of therapy within everyday communication is the holy grail of aphasia treatment design and evaluation. Whilst impaired sentence production is a predominant symptom of Broca's-type aphasia, the effects of sentence production therapy on everyday conversation have not been investigated. Given the context-sensitive nature of spoken production by people with aphasia, it is difficult to extrapolate implications for everyday conversation based on results from task-based assessment (such as picture description, story retell or interview). Thus, there are strong theoretical and clinical motivations to investigate generalization from sentence production treatment to everyday conversation. Aims: To evaluate a theoretically driven treatment focused on the language production skills of participants with post-stroke Broca's aphasia and to track outcomes from psycholinguistic assessment tasks to everyday conversation. Methods & Procedures: A case series design was utilized with pragmatic selection of participants with chronic aphasia undergoing the same assessment and treatment procedures. Nine participants with Broca's aphasia and their main conversation partners took part in the study. Treatment was implemented once weekly over 8 weeks and targeted production of basic syntax--two, three and four constituent constructions--through principles of mapping and reduced syntax treatment. Use of different possible exemplars for nouns, particularly pronouns, was trained together with use of both light and heavy verbs. Participants had the opportunity to "top-up" therapy practise by completely a homework task that mirrored the therapy task. Outcomes & Results: Syntactic well-formedness was assessed in samples of constrained sentence production, narrative retell and naturally occurring conversations at baseline, 1 week post-treatment, and 1 month post-treatment. Treatment showed strong direct effects in trained and untrained sentence construction tasks, with some generalization to narrative retell tasks. There was little evidence of change in everyday conversation. Conclusions & Implications: Improvement in language production in constrained assessment tasks may not impact on everyday conversations. Implications for further research are discussed, e.g. the need for bridging interventions between constrained and unconstrained contexts of language production. Clinical implications include the potential to streamline therapy planning and delivery by making use of rich, hybrid therapies to treat individuals with similar symptom profiles but with a range of underlying deficits.