In order to improve our understanding of the role of Clostridium difficile in infants we characterised the strains isolated from this population. The production of toxin A and toxin B was studied. The toxin A, playing a major role in the disease, was searched for in faecal samples. The serogroup of the isolates was determined because some serogroups have been shown to be more pathogenic than others. Over a 9-month period, 102 faecal samples from 102 hospitalised infants (0-12 months) were analysed and 26% of the children were colonised with C. difficile. Fifteen isolates secreted neither toxin A nor B (62.5%). Nine isolates were toxigenic and secreted both toxins (37.5%). Of the eight toxigenic strains tested, six were from serogroup H and two serogroup K. Of the 13 nontoxigenic strains tested, 8 belonged to serogroup D, 2 to serogroup X, and 1 each to serogroup A, serogroup B and serogroup C. Three infants out of 102 studied had toxin A in their faeces. In summary, the infants can be colonised by (1) nontoxigenic strains, most of them from nonpathogenic serogroup D, without toxin A in the faeces; (2) toxigenic strains of virulent serogroups H and K, with or without toxin A in the faeces. Although some infants had diarrhoea, none needed a specific treatment for C. difficile. No specific C. difficile pathology could be retained and different mechanisms are advanced to explain this absence of pathogenicity.