We first determined the analytical specificity and ubiquity (i.e., the ability to detect all or most strains) of a Clostridium perfringens-specific real-time PCR (rtPCR) assay based on the cpa gene (cpa rtPCR) by using a bacterial strain panel composed of C. perfringens and non-C. perfringens Clostridium strains. All non-C. perfringens Clostridium strains tested negative, whereas all C. perfringens strains tested positive with the cpa rtPCR, for an analytical specificity and ubiquity of 100%. The cpa rtPCR assay was then used to confirm the identity of 116 putative C. perfringens isolates recovered after filtration of water samples and culture on mCP agar. Colonies presenting discordant results between the phenotype on mCP agar and cpa rtPCR were identified by sequencing the 16S rRNA and cpa genes. Four mCP(-)/rtPCR(+) colonies were identified as C. perfringens, whereas 3 mCP(+)/rtPCR(-) colonies were identified as non-C. perfringens. The cpa rtPCR was negative with all 51 non-C. perfringens strains and positive with 64 of 65 C. perfringens strains. Finally, we compared mCP agar and a CRENAME (concentration and recovery of microbial particles, extraction of nucleic acids, and molecular enrichment) procedure plus cpa rtPCR (CRENAME + cpa rtPCR) for their abilities to detect C. perfringens spores in drinking water. CRENAME + cpa rtPCR detected as few as one C. perfringens CFU per 100 ml of drinking water sample in less than 5 h, whereas mCP agar took at least 25 h to deliver results. CRENAME + cpa rtPCR also allows the simultaneous and sensitive detection of Escherichia coli and C. perfringens from the same potable water sample. In itself, it could be used to assess the public health risk posed by drinking water potentially contaminated with pathogens more resistant to disinfection.