The primary objective of this paper is to review theoretical and methodological literature pertaining to the clinical evaluation of discourse abilities in speakers who have sustained traumatic brain injury (TBI). A brief history of the study of discourse impairment in this population is followed by consideration of the following issues: (1) sampling (genres selected, the physical setting in which sampling takes place, the relationship between speakers, elicitation techniques, presence of recording devices, the number of samples required, and transcription); (2) measurement; (3) the relationship between sampling and measurement; (4) other approaches to discourse assessment (self and close other report); (5) consideration of the criterion of 'normal' which clinicians should employ; (6) the relationship between discourse impairment and measures of executive function; and (7) the relationship between discourse impairment and seventy of injury. Recommendations arising from a critical review of these domains are made for both clinical practice and research.