The first Dutch outbreak due to Clostridium difficile ribotype 027 was observed in mid-2005; by the end of that year, eight hospitals were affected. To study the relationship between hospital-wide antibiotic use and the incidence of 027-linked C. difficile-associated disease (CDAD) three study groups were made: group A, all eight hospitals with an 027-associated epidemic; group B, five of a total of six hospitals with occasional 027 cases, without an increase in CDAD; and group C, ten randomly selected hospitals with no reported 027 epidemics or isolated 027 cases. Quarterly data on CDAD incidences, hygiene measures and the use of fluoroquinolones, second- and third-generation cephalosporins, extended-spectrum penicillins, penicillins with beta-lactamase inhibitors, carbapenems, lincomycins and macrolides were collected for 2004 and 2005, and divided into pre-epidemic and epidemic periods. Using a multilevel Poisson regression analysis, CDAD incidence was linked to antibiotic use in the previous quarter and to certain hygiene measures. In the pre-epidemic period, the total use of the studied antibiotics was comparable between affected and unaffected hospitals. Higher use of second-generation cephalosporins, macrolides and all of the studied antibiotics were independently associated with a small increase in CDAD incidence [relative risk (95% confidence interval): 1.14 per increase of 100 defined daily doses per 10,000 bed days (1.06-1.23), 1.10 (1.01-1.19) and 1.02 (1.01-1.03), respectively]. However the effect was too small to predict which hospitals might be more prone to 027-associated outbreaks.