The incidence of Clostridium difficile-associated diarrhoea (CDAD) was investigated retrospectively at a 690-bed teaching hospital for the period 1983-92. Our aims were to determine: (i) the distribution by age and sex of patients with CDAD, (ii) the possibility of a seasonal trend and, (iii) the influence of infection control procedures, contamination of the hospital environment and the use of third-generation cephalosporins. The laboratory diagnosis of CDAD was based on demonstration of the organism by stool culture and/or detection of specific cytotoxin in stool filtrates. C. difficile was detected in 917 patients who were being investigated for diarrhoeal illness. Yearly isolations varied from a low of 49 in 1983 to a high of 120 in 1990 (Chi square for linear trend 128.8; P < 0.005). Most patients were elderly, with 63% aged 60 years or more; the majority (59%) were female. The relationship between culture of C. difficile and detection of cytotoxin in faecal extracts was also examined. Sixty percent of a sample of 132 isolates from patients in whom faecal cytotoxin was not detected produced cytotoxin in vitro, suggesting that culture is a more sensitive indicator of infection with C. difficile than cytotoxin detection. When the total number of faecal specimens received in the laboratory was used as a denominator there was an increase in the number of incident cases of CDAD between 1983 and 1990, apart from 1986. When occupied bed days was used as the denominator a similar trend was observed with a peak in 1990.(ABSTRACT TRUNCATED AT 250 WORDS)