BACKGROUND: Anomia therapies can be broadly categorised into semantically based and word-form based. Comparable treatment outcomes of these different approaches have been reported. Law, Wong, Sung, and Hon (2006a) have recently applied semantic feature analysis (Boyle and Coelho, 1995) on three Chinese anomic speakers and found treatment progress limited to those with relatively mild semantic deficits. Hence the present study explored alternative approaches to manage naming problems. Despite the challenges to the use of cueing that stem from the characteristics of the Chinese script -- i.e., the lack of mapping between sub-character components and phonemes -- an intervention was designed in which participants were trained to associate objects with letters representing the initial consonants of the target names, from which phonological cues were generated to assist in word retrieval. AIMS: This study evaluated the effectiveness of an ortho-phonological therapy for Chinese anomia. Given its function of strengthening semantic and phonological representations, the therapy was predicted to have item-specific treatment effects. METHODS AND PROCEDURES: Four Cantonese-speaking anomic individuals participated in the study. Two were hypothesised to have disrupted access from semantics to phonology, whereas the other two suffer additional impairment to semantic processing and phonological output. A multiple baseline design was adopted, consisting of pre-treatment training in which participants learned to generate CV syllables in response to letter cues, then a baseline, one or two treatment phases, and a maintenance phase. A cueing hierarchy was used during therapy, which began with a letter cue, followed by the provision of increasing phonological information if the participant continuously failed to retrieve the target name. OUTCOMES AND RESULTS: All participants benefited from the intervention and could maintain treatment gains for at least 1 month. Item-specific improvement was observed for two participants, whereas the third patient also made progress on some of the untrained items that shared the same initials as the treated stimuli, and treatment generalisation of the fourth patient further extended to items with initial consonants different from the trained items. Different levels of facility in employing letter-sound correspondence during naming and executive problem-solving skills were argued to have contributed to discrepant treatment outcomes. CONCLUSIONS: The ortho-phonological therapy is a viable option for rehabilitating anomia in Chinese. It has the potential for being a naming strategy that may result in general improvement in word retrieval. Our findings echo the view in recent studies that urges investigation into the role of non-linguistic cognitive abilities in predicting treatment outcomes.