This paper presents a single case treatment study of a man with dysphasia following stroke. The aim of the treatment was to improve his severely impaired word-finding abilities. Three interventions are described: a pilot study, a semantic and lexical treatment study, and treatment with a cueing aid. The pilot study demonstrates the effectiveness of a written lexical task in improving spoken naming. The semantic and lexical treatments give results which contrast sharply with the result of the pilot study and emphasise the importance of looking at exactly how treatment tasks are carried out when evaluating their effects. Treatment with the cueing aid resulted in a highly significant improvement which generalised to naming of untreated items and was maintained for over 15 months after the treatment was completed. An improvement in connected speech was also demonstrated. The implications for theory and clinical practice are discussed.