Two hundred obligately anaerobic bacterial isolates from human clinical sources were tested for susceptibility to ceftizoxime using a standard National Committee for Clinical Laboratory Standards microwell system. In general, the minimal inhibitory concentrations (MIC50) ranged from less than 0.125 to 32 microgram/ml; however, the MIC50 for Bacteroides fragilis averaged greater than 128 microgram/ml. This finding is inconsistent with the results of in vivo testing of ceftizoxime in an animal model of intra-abdominal sepsis (B fragilis is a major contributor to the development of intra-abdominal abscesses). Various modifications of in vitro assay parameters, including basal media (brain-heart infusion [BHI] or Wilkins-Chalgren [WC]) and methods (microwell, broth, and agar dilutions), were compared. Ten B fragilis isolates from the original clinical study were used. Results indicate that the activity of ceftizoxime decreased between two- and fourfold after storage for 48 hours at -80 degrees C, regardless of methodology or basal media. When microwell was compared with broth dilution, there was a four- to 64-fold decrease in MIC values by the latter method using BHI but little variation using WC. No differences were observed when the incubation time was varied. Preliminary data indicate that MIC values from broth dilution using BHI correspond with those of agar dilution assays. These results suggest methodologic as well as environmental discrepancies with regard to susceptibility testing of ceftizoxime. These differences may lead to misinterpretation of the true susceptibility of organisms to this agent, particularly when the results are compared with in vivo observations.