We compared the susceptibility results for 200 clinical anaerobes with nine antibiotics obtained by using a new ATB ANA (bioMérieux) device against those obtained by the National Committee for Clinical Laboratory Standards (NCCLS) standard agar dilution method. For better evaluation of the device, we added some resistant Bacteroides fragilis group strains from our own collection: 3, 6, and 12 strains that were resistant to imipenem, ticarcillin plus clavulanic acid, and co-amoxiclav, respectively, and 2 other strains with decreased susceptibility to metronidazole. For some strains that did not grow on ATB S medium, tests were performed by using West-Wilkins medium supplemented with 1.5% agar. The new ATB ANA device made clinical categorization of the investigated strains possible, according to French (Committee of the Antibiogram of the French Society of Microbiology) or U.S. (NCCLS) breakpoints, with the following respective results: category agreement, 94.3 and 94.9%; minor errors, 4.8 and 3.8%; major errors, 0.4 and 0.8%; and very major errors 4.6 and 4.2%. The ATB ANA device was able to detect low-level metronidazole-resistant B. fragilis strains according to the French breakpoints but not the NCCLS ones. For B. fragilis and -lactamase-positive Prevotella strains, the clustering effect of amoxicillin MICs around the French breakpoints led to more frequent minor errors. ATB ANA is a very convenient method to determine the antibiotic susceptibilities of anaerobes. Results obtained by ATB ANA correlated well with those obtained by the reference method.