Agar minimal inhibitory concentrations and mouse protection test effective doses were determined for each of four beta-lactam antibiotics against each of 12 Gram-negative and 3 Gram-positive bacterial cultures. The beta-lactamase activity of these cultures also was studied. The data were examined to determine whether relative in vivo efficacies could be predicted from relative in vitro activities. Although such predictions were quite accurate for cefoxitin and cefazolin, this was not true for cefamandole or for cephalothin. Such poor predictability was not necessarily associated with the susceptibility of these cephalosporins to hydrolysis by bacterial beta-lactamases. Although the clinical significance of these observations is not known, these data emphasize that relative in vitro activities should be used only with caution to estimate in vivo efficacies, since not all compounds show the excellent predictability observed here for cefazolin and cefoxitin.