Background: Impairments of word retrieval and production are a common and distressing feature of aphasia, and much clinical time is devoted to attempts at their remediation. There are now many research papers devoted to case studies examining treatments for word-retrieval impairments using a wide range of tasks with individuals who have varying levels of impairment. Aims: This paper aims to continue the selective review of this literature carried out by Nickels and Best (1996a). It summarises in table form those published papers since 1980 which present single case studies of treatment for word-retrieval impairments and which satisfy minimal methodological criteria. Main contribution: Several main themes are derived from the literature and discussed in more detail, these include strategic approaches and facilitative or repair approaches to remediation, the contrast between semantic and phonological tasks in therapy, generalisation in therapy tasks and the relationship between impairment, therapy task, and outcome. Further discussion relates to the relationship between impairment level treatments, and measures of disability and handicap, and between therapy research and therapy practice. Conclusions: There are now many research papers devoted to impairments of word retrieval, and there can be no doubt that therapy for word-retrieval impairments can be highly successful, resulting in long-term improvements which can be of great communicative significance for the individual with aphasia. However, predicting the precise result of a specific treatment task with a specific individual with certainty is still not possible. For clinicians the recommendation is to use analyses of functional impairments to guide the choice of task, but to ensure that efficacy is tested and not assumed. Furthermore, structured multi-modal and multicomponent tasks (e.g., "semantic" or "phonological" cueing hierarchies) may hold the most promise for many individuals. For researchers, there remains a need to further dissect tasks, impairments, and their interactions across series of single cases.