The impact of changes in antibiotic policy on Clostridium difficile-associated diarrhoea (CDAD), over a five-year period between 1995 and 2000, were studied in the Preston Acute Hospitals Trust. In 1996 the policy was changed in the Preston Acute Hospitals Trust from cefotaxime to ceftriaxone for initial treatment of severe sepsis or pneumonia in medical patients. Over the next nine months the average number of patients with C. difficile toxin-positive stools per quarter increased from 16 to 39. The predicted use of ceftriaxone exceeded by 65% an estimate based on prior use of cefotaxime. A policy of restricted duration of ceftriaxone was introduced, and although this reduced usage by over 50%, CDAD continued at an average of 9.2 cases per month, despite withdrawal of oral cephalosporins in December 1998. In August 1999 levofloxacin was substituted for ceftriaxone in the policy. The incidence of CDAD fell progressively to five cases per month by 2000. It would appear that a short (typically three dose) course of third-generation cephalosporin poses a similar risk for CDAD as a more prolonged course. The six-month delay in the decline of CDAD after virtual withdrawal of cephalosporins may reflect a slowly diminishing environmental reservoir.