In vitro susceptibility testing is only one step in the evaluation of the potential efficacy of antimicrobial agents against the Bacteroides fragilis group. An assessment of in vivo efficacy, with a consideration of the factors that can best be studied in an infected host, is also an integral part of this process. Abscess models in rodents have been used to correlate in vitro activity with in vivo efficacy against this group of microorganisms. For metronidazole, clindamycin, moxalactam, and cefoxitin, the correlation was strong; for chloramphenicol and carbenicillin, it was not. In vivo studies of mixed infection with the B. fragilis group and Escherichia coli showed that cefoxitin and imipenem were effective; in contrast, cefotetan was not effective against resistant strains. Only strains susceptible to ceftizoxime in the agar dilution test were also affected by this drug in vivo. The so-called inoculum effect noted with ceftizoxime may explain this finding. In vivo elimination of encapsulated organisms of the B. fragilis group was found to be more difficult than elimination of unencapsulated isolates. The beta-lactamase produced by Bacteroides species can protect the enzyme-producing organism as well as its partners in mixed infections from the effects of beta-lactam antibiotics. These data illustrate the complexity and difficulties encountered when in vitro activity is correlated with in vivo efficacy.