Clostridium difficile diarrhoea represents a significant health-service burden. We recently experienced an outbreak of C difficile diarrhoea associated with increased use of cefotaxime. The question we pose in this paper is how did the introduction and withdrawal of a single antibiotic so greatly affect rates of C difficile diarrhoea? Other antibiotics had nearly as high a risk of causing diarrhoea as cefotaxime, and the majority of patients never received cefotaxime. We believe that such outbreaks of C difficile diarrhoea are best understood in terms of a population model, and that taking antibiotics like cefotaxime should be thought of as a population rather than an individual risk factor. We postulate a herd-immunity model of C difficile diarrhoea, and examine the implications of this hypothesis.