A rapid and sensitive procedure for estimating low numbers of Clostridium perfringens has been investigated and compared to methods used currently in the food industry. The new liquid medium, RPM (rapid perfringens medium), was compared with sulfite-polymyxin-sulfadiazine agar and tryptose-sulfite-cycloserine agar in recovery studies with naturally contaminated and with inoculated foods. The medium consists of a mixture of litmus milk and fluid thioglycolate medium fortified with glucose, peptone, gelatin, yeast extract, sodium chloride, and ferrous sulfate. Selectivity is based on an antibiotic system (polymyxin B sulfate and neomycin sulfate) incorporated into the medium, coupled with an incubation temprature of 46 to 48 degrees C for 24 h. Tubes were scored as positive if a stormy fermentation was observed. All tubes demonstrating stormy fermentation were confirmed as containing C. perfringens. Of a total of 774 naturally contaminated food samples, 546 samples (71%) were found to contain C. perfringens with RPM, whereas only 168 (22%) of the samples were positive using sulfite-polymyxin-sulfadiazine agar. C. perfringens was isolated from 71% of 85 other samples using RPM as compared to 14% with tryptose-sulfite-cycloserine agar. Enumeration studies on 14 individual samples using the most probable number technique also demonstrated greater sensitivity with RPM.