In a cross-sectional survey of 226 HIV-infected men, we examined the occurrence of diarrhoea and its relationship to drinking water consumption patterns, risk behaviours, immune status and medication use. Diarrhoea was reported by 47% of the respondents. Neither drinking boiled nor filtered water was significantly associated with diarrhoea (OR = 0·5 [0·2, 1·6], 1·2 [0·6, 2·5] respectively), whereas those that drank bottled water were at risk for diarrhoea (OR = 3·0 [1·1, 7·8]). Overall, 47% always or often used at least one water treatment. Of the 37% who were very concerned about drinking water, 62% had diarrhoea, 70% always or often used at least one water treatment. An increase in CD4 count was protective only for those with a low risk of diarrhoea associated with medication (OR = 0·6 [0·5, 0·9]). A 30% attributable risk to diarrhoea was estimated for those with high medication risk compared to those with low medication risk. The significant association between concern with drinking water and diarrhoea as well as between concern with drinking water and water treatment suggests awareness that drinking water is a potential transmission pathway for diarrhoeal disease. At the same time we found that a significant portion of diarrhoea was associated with other sources not related to drinking water such as medication usage.