At the present time, there is no convincing indication that Crohn's disease is a bacterial disease, although an association with mycobacteria has been hypothesised for many years. The hypothesis that bacteria could be the cause, or at least an important concause of Crohn's disease is supported by several experimental and clinical observations: animals kept in a germ-free environment fail to develop intestinal inflammation; bacteria are the cause of human and animal intestinal diseases similar to Crohn's disease; luminal content is necessary for causing gut lesions; and, moreover, antibiotics are successfully used in the treatment of Crohn's disease. Bradford Hill criteria recently used to assess a causal relationship for Helicobacter pylori and peptic ulcer can be applied for establishing or excluding a causality between mycobacteria and Crohn's disease. Of these criteria, only biological plausibility, coherence and analogy are satisfied. However, failure to identify a specific pathogen does not exclude a possible role for bacteria in causing Crohn's disease lesions and symptoms. Pathogenic or commensal enteric bacteria could overinfect the primary lesions, leading to chronic intestinal inflammation in genetically susceptible hosts. Another possibility is that components of the normal intestinal flora could acquire pathogenic characteristics.