BACKGROUND: Repetition priming is often a component of treatments for word-finding disorders. It can facilitate or interfere with naming success depending on a number of factors. Here we investigate the effectiveness of massed priming coupled with semantic or phonological context as a treatment for naming impairments arising from semantic and phonological deficits. AIMS: We aimed to determine whether (1) this procedure, used previously in a short-term facilitation study, would effectively improve word retrieval in a treatment study, and (2) the pattern of facilitation or interference observed in the facilitation study would carry over to the treatment programme. METHODS AND PROCEDURES: We used a single subject multiple baseline design. There were two participants: LP with a phonological encoding deficit and AS with both semantic and phonological deficits. Treatment involved identifying and repeating the names of words that were related semantically or phonologically, or unrelated. Pre and post measures of naming were used to assess overall effectiveness of the treatment. Acquisition, maintenance, and generalisation were measured with baseline tests at the start of each session. Correct responses and errors on within-training naming probes were used to measure sensitivity to priming in a particular context. OUTCOMES AND RESULTS: LP benefited from this procedure regardless of the training context. AS showed interference in the semantic context during training and only modest short-term gains. These outcomes were predicted by their performance on an earlier facilitation study. CONCLUSIONS: Contextual repetition priming has different effects on naming and these differences appear to be related to the context of training (semantic or phonological) and the primary source of an individual's naming impairment (semantic or phonological). This procedure is most effective when semantic processing of words is relatively spared.