BACKGROUND: The investigation into conversational interaction in aphasia and its therapeutic implications has received much interest. It is increasingly accepted that people with aphasia cannot be treated in isolation but as part of a social unit. Training conversation partners has been shown to impact on the disabling effect of aphasia. AIMS: The principal aim of this study was to evaluate a programme designed to train a relative/friend to communicate more effectively with an aphasic partner. The intervention was assessed in terms of effects on communication and on wellbeing. METHODS AND PROCEDURES: Four single-case studies were completed. The aphasic participants had severe impairment affecting both comprehension and expression. The baseline was a series of three assessments. The dyads were videotaped having a conversation, and completed a Visual Assessment for Self-Esteem Scale (VASES). The non-aphasic partner also completed a Hospital Anxiety and Depression Scale (HADS). The training programme ran for 5 weeks (1.5 hours/week) at the couples' homes. It included education, video feedback, and role-play. The initial assessments were repeated post-intervention. OUTCOMES AND RESULTS: Conversation data were subject to a conversation analysis as well as frequency counts of nonverbal behaviours. We found positive individual changes and positive trends: three of the four dyads increased their use of gesture and the proportion of successful repair sequences increased post-intervention, although these changes were not statistically significant. There were no significant differences for the HADS but there were differences between the aphasics and the non-aphasics on the VASES. CONCLUSIONS: The positive trends suggest that working with couples can produce beneficial effects. Future research may help establish who will benefit from this approach and the optimal number of sessions.