The purpose of this study was to find out the actual circumstances which lead to dysentery-related deaths in the rural Bangladesh. The Community Health Workers of the International Centre for Diarrhoeal Disease Research, Bangladesh have been collecting records of all deaths at the rural area of Matlab through a surveillance system since 1963. A review of existing data on dysentery-related deaths from this area during 1976-1981 suggests that deaths in children followed a recurring seasonal pattern with an increase during the post-monsoon season of August-November of each year. This seasonal pattern of death was not evident among adults. The overall dysenteric death rate during 1978-1981 was 13.3 per 10,000 population per year. The highest rates were in patients of the two extreme age groups. Deaths reported recently by the health workers were re-investigated. Although the causal agents producing fatal dysentery in most patients in the community remained unidentified, it was likely to be species of Shigella in childhood deaths. To identify clinical determinants of a fatal outcome, a case-control analysis was done with patients hospitalised with dysentery in 1980. The risk factors shown to be significantly associated with deaths were: longer median duration of illness (p = less than 0.001), female sex (p = 0.039), signs of respiratory infection (p = less than 0.001) and severe malnutrition (p = 0.002).