The in vitro susceptibility to cefamandole, cefoxitin, and cefuroxime of clinical isolates of Streptococcus faecalis, Klebsiella, and indole-positive Proteus was assayed using minimum inhibitory concentration (MIC) determinations in broth and on solid media. It could be demonstrated that the agar dilution MICs obtained when S. faecalis was tested against cefuroxime and when indole-positive Proteus strains were tested against cefamandole tended to be considerably lower than those obtained with the broth dilution technique. The Klebsiella strains tested did not show any major differences with regard to MICs in broth or on solid media. Using an animal experimental infection model it could be demonstrated that with indole-positive Proteus the higher broth MIC correlated better to the observed data than the lower agar MIC when a beta-lactamase-producing strain was tested. The data obtained indicated that the beta-lactamase of the indole-positive Proteus strain was inducible. The results of the study gave evidence for a risk of false susceptibility of some bacterial species against cefamandole when agar techniques, e.g. paper disk diffusion, are used. For cefuroxime, the same phenomenon can be expected with S. faecalis and to some extent with indole-positive Proteus. In this study, cefoxitin seemed considerably less affected by the technique used for susceptibility testing.