BACKGROUND: Computer-based rehabilitation programs are now available for patients' use at home and in the clinical setting, yet we have meagre outcome data associated with their usefulness under self- and/or clinician-guided conditions. Aims: We assess the benefits of a computer-delivered, hierarchical phonological cueing protocol (cued naming) under two conditions of instruction, (1) with full clinician guidance or (2) in partial independence. METHODS AND PROCEDURES: We employed a single-subject experimental design, which was replicated over six chronic aphasic subjects, three in each instruction condition. Subjects with deficits identified as primarily phonological in nature were administered a phonological treatment, utilising a computerised therapy program (MossTalk Words), under one of the two conditions. OUTCOMES AND RESULTS: Training-specific acquisition and maintenance was demonstrated in both conditions. Limited and variable generalisation patterns were noted. CONCLUSIONS: Chronic aphasic individuals with moderate-to-severe phonologically based naming impairment can benefit from a computerised cued-naming protocol and independent work on the computer can be an effective adjunct to clinician-guided therapy.