Background: An individual showing an intricate pattern of errors as a consequence of brain damage affords the clinician an opportunity to analyse language data, the exact pattern of which would be impossible to generate. This situation is particularly valuable when the phenomenon to be investigated is unique. This is the case with verbal perseveration. It does not always occur in aphasia. However, when it does, qualitative analysis of the perseveratory errors sheds light on the individual's underlying disorder(s). The language data reported on in this paper are discussed in detail in German in my unpublished PhD thesis (Stark, 1984). I would like to express special thanks to Karin Geisler, Leena Martin, Christiane Pons, and Heinz Karl Stark for their constructive comments and assistance on versions of this paper. I would also like to thank the reviewers of this paper for their valuable critique and suggestions. Aims: In this paper verbal perseverative errors produced by a client with transcortical sensory aphasia during routine language testing are analysed. The aim of this study is to characterise the types of perseverative responses and related phenomena and to arrive at an understanding of the mechanisms of the underlying disorders on the various linguistic levels. Methods & Procedures: Client MH, clinically diagnosed as having transcortical sensory aphasia, was administered several language tests. The language data were transcribed and a qualitative analysis of her oral language production was performed with particular emphasis on perseverative and paraphasic responses of the semantic, syntactic, and ideational error types. Outcomes & Results: Verbal perseveration is the most prominent symptom in client MH's language data. The perseveratory errors display a complex development within a task (interstimuli), across tasks and task types and they extend over 2 days. The produced perseverations in the different tasks vary in their linguistic complexity from word‐level to text‐level responses. Analysis of MH's perseverative errors within the framework of Levelt's (1993) model of language production and comprehension with modifications from Dietrich (2002) reveals impairment to several processing components of the conceptualiser: monitoring, discourse processing, message generation, and lexical/semantic aspects of the lexicon. Conclusions: The pattern of perseveratory errors in client MH lends support to the relative preservation of grammatical encoding, phonological encoding, and articulation. She produced well‐articulated, grammatically correct utterances. However, her responses were semantically inadequate and she showed word retrieval deficits. Pre‐ and post‐production monitoring was impaired for semantic aspects of the utterances. Few phonological errors were made, and they were immediately corrected by MH. Self‐corrections were not observed for semantically inadequate and contradictory responses.