BACKGROUND: Based on models of speech perception and production, different stages in phonological processing can be distinguished: phonetic, sublexical, and lexical levels of decoding and encoding respectively. In theory they can be selectively impaired. To address the specific level of phonological processing disruption we developed training material that entailed practice demands for each level of processing. AIMS: This paper aims to demonstrate the utility of this material-based approach by presenting the results of PS, a 52-year-old man with chronic conduction aphasia who showed both decoding and encoding difficulties at sublexical levels. METHODS & PROCEDURES: The main component of the treatment programme was practice of minimal phonemic contrasts. The training material consisted of monosyllabic stimuli that were systematically varied for certain linguistic criteria, such as phonemic contrast position (onset/coda), lexicality (words/pseudowords), and phonetic complexity (increasing/unspecific sonority). The impact of these factors was studied in an alternating-treatments design employing control tests to assess baseline, outcome, and maintenance. The treatment lasted for 6 weeks during which phonological processing was practised under six treatment conditions. All exercises were computer assisted, and each session consisted of three main tasks: discrimination, identification, and reproduction. OUTCOMES & RESULTS: As predicted, the participant showed improvement during therapy when practising those items that called specifically for sublexical phonological processing. CONCLUSIONS: This study demonstrates that an impairment-specific and material-based therapy approach is promising for the treatment of impaired sublexical phonological processing. Furthermore, the results suggest common mechanisms of input and output phonology.