The effect on diagnostic yield of testing sequential stools was assessed during two hospital epidemics of Clostridium difficile. Using a rapid immunoassay, C. difficile-associated disease was diagnosed in 237 diarrhoeal patients, of whom 204 (86%) were diagnosed from the first faeces sample and 12 (5%) were diagnosed from follow-up samples obtained within 1 week. The remaining 21 (9%) patients yielded a positive test from stools obtained >1 week after the initial negative sample. It was concluded that repeated testing of stools for C. difficile toxin is of value in controlling outbreaks of C. difficile infection.