Susceptibility testing of 161 clinical isolates of the Bacteroides fragilis group was performed to compare interpretive results generated by the broth disk elution and broth microdilution methods recommended by the National Committee for Clinical Laboratory Standards. Among the cephalosporin-cephamycin compounds tested, correlation was poorest for ceftizoxime (71%), ceftriaxone (57%), and cefotaxime (47%); when the tests did not correlate, false resistance was seen 92, 95, and 93% of the time, respectively. Cefotetan and cefoperazone showed lack of correlation in 19 and 20% of the tests, respectively. For cefotetan, false resistance was more frequent, while with cefoperazone, false susceptibility occurred more often. Cefoxitin produced the fewest discrepancies; 10% of the disk elution tests produced either false-resistance or false-susceptibility results. Mezlocillin and piperacillin showed lack of correlation in 8 and 14% of the tests, respectively, and discrepancies were due primarily to false-resistance results. Overall with the beta-lactams, 84% of the discordant interpretive results were false resistance by the broth disk elution test. Clindamycin had a discrepancy rate of 10%, with the majority of discrepancies being false susceptibility disk elution results. Because of the high number of discrepancies noted with ceftizoxime, ceftriaxone, and cefotaxime, we recommend that these drugs not be tested by the disk elution method and that they be tested by a quantitative MIC method such as the broth microdilution test. Furthermore, caution should be exercised when interpreting broth disk elution results with all the beta-lactams included in this study except imipenem. These data indicate the lack of correlation of results between these two tests for many beta-lactams and suggest the need for a reexamination of the disk elution method to provide a more accurately standardized test.