Two closed-head-injured subjects and 12 non-brain-damaged controls explained a novel procedure to a blindfolded third person. Their productions were transcribed and scored by nine judges on ratings scales based on Grice's maxims of quantity and manner. The two closed-head injury (CHI) productions were significantly inferior to the controls, both being rated as disorganized, confusing, and ineffective. While one subject's production was considered overly repetitive, the other had too little detail. In order to specify, more exactly, the nature of the deficits, a cohesion analysis and an analysis of informational content were performed. The cohesion analysis suggested that the CHI productions were generally similar to the controls. The informational analysis was more revealing. Judgements of detail and repetitiveness were reflected in the number of new and repeated propositions. Errors in sequencing and inclusion of irrelevant propositions contributed to the disorganized and confusing nature of the texts. The results were interpreted within the framework of frontal lobe deficits in monitoring and regulation of performance.