Recent clinical isolates (approximately 150 strains) of the family Enterobacteriaceae were examined by agar diffusion, microdilution, and the Autobac automated system for their responses to beta-lactam antibiotics singly and in combination with amdinocillin (formerly called mecillinam). The ratio of ampicillin, carbenicillin, and cephalothin to amdinocillin was maintained at a 10:1 ratio in most of the evaluations. The same isolates were studied in mice challenged with 100 to 1000 LD50s and treated with graded doses of the antibiotics singly and in combination. Efficacy in vivo was based on the concentration of antibiotic in milligrams per kilograms (mg/kg) required to protect 50% of the animals (PD50). After a single administration of the antibiotics, plasma levels were determined in the critical time period (30 min to 4 hr) during which the acute, overwhelming systemic infections could be controlled by appropriate therapy. Regression curves comparing in vivo and in vitro results were used to establish cut-off points for categorizing bacterial susceptibility in each of the laboratory tests for the single agents and combinations. A high degree of synergism between amdinocillin and the beta-lactam agents was demonstrated in animals (54 to 78% of the strains examined) and to a lesser extent by laboratory methodologies. There was an excellent correlation of in vivo and in vitro responses to ampicillin, carbenicillin, and cephalothin alone and in combination with amdinocillin for those species for which the single antibiotics are generally indicated. The correlations validated the chosen cut-off points. The correlation of in vivo and in vitro responses to the single or combined antibiotics was generally poorer for those species not usually responsive to the single antibiotics. The greatest difficulty in predicting proper in vivo responses, based on the results of in vitro tests, was observed with amdinocillin.