Despite significant progress in the psycholinguistic study of translation, research on its neurological underpinnings has been limited and sparse. Translation scholars have recently taken an interest in relevant neuroscientific evidence, focusing on imaging studies. This paper addresses the issue by considering an equally important body of data: clinical evidence. Specifically, a hypothesis-driven analysis is offered of 21 cases of brain-lesioned bilinguals exhibiting translation disorders. Three neurofunctional and three neuroana-tomical hypotheses are derived from the Revised Hierarchical Model and the Declarative/Procedural Model, respectively. Consistent with relevant predictions, the evidence suggests that there are neurofunctionally independent routes for translation, as opposed to monolingual speech production; backward, as opposed to forward, translation; and form-based, as opposed to conceptually mediated, translation. Available data further indicates that word and sentence translation are critically subserved by posterior brain areas implicated in declarative memory, and by frontobasal areas implicated in procedural memory, respectively. In addition, translation routes appear to be entirely left-lateralized.