Many media have been developed to enumerate Clostridium perfringens from foods. In this study, six media [iron sulfite (IS) agar, tryptose sulfite cycloserine (TSC) agar, Shahidi Ferguson perfringens (SFP) agar, sulfite cycloserine azide (SCA), differential clostridial agar (DCA), and oleandomycin polymyxin sulfadiazine perfringens (OPSP) agar] were compared in a prestudy, of which four (IS, TSC, SCA, and DCA) were selected for an international collaborative trial. Recovery of 15 pure strains was tested in the prestudy and recovery of one strain from foodstuffs was tested in the collaborative trial. Results from the prestudy did reveal statistical difference of the media but recoveries on all media were within the microbiological limits (+/- 30%) of IS, which was set as a reference medium. Recoveries on the media tested in the collaborative trial were statistically different as well, but these differences were of no microbiological-analytical relevance. Food matrices did not affect the recovery of C perfringens in general. DCA and SCA, in particular, are labor-intensive to prepare and DCA frequently failed to produce black colonies; gray colonies were quite common. Since IS medium is nonselective, it was concluded that TSC was the most favorable medium for the enumeration of C. perfringens from foods. (C) 2003 Elsevier Science B.V. All rights reserved.