There has been a great deal of speculation regarding the role of microcomputers in the retraining of dysphasic patients. Enthusiasm should be tempered with caution and each application in this new and exciting area has to be clearly defined if it is to have an impact in the future. This study is one of a series and compares the performances of dysphasic stroke patients on selected speech therapy tasks administered both in the conventional way and using a microcomputer. The twelve subjects were assessed on linguistic and non-linguistic tasks, which were matched across the two situations. Both quantitative and qualitative measures of patients performances were analysed. After an initial familiarity task, no significant differences were noted between the scores in each situation. Theoretical and practical implications of using the computer to supplement speech therapy are discussed.