Rabbits with Staphylococcus aureus endocarditis were treated with cephalothin or ceftazidime to determine whether differences in in vitro activity would result in differences in in vivo efficacy. Antibiotics were administered in doses equivalent to maximum recommended human doses, and results of laboratory tests to predict antimicrobial efficacy were determined during treatment. Cephalothin and ceftazidime MICs for the challenge strain were 0.5 and 8 ,ug/ml, respectively. MBCs were 32 and >128 ,ug/ml, respectively. With peak sera, laboratory results (means) for cephalothin and ceftazidime were as follows: ratios of concentration in serum to MIC, 300 and 16; ratios of concentration in serum to MBC, 4.8 and <1; bacteriostatic antibacterial activity titers in serum, 1:256 and 1:16; and bactericidal antibacterial activity titers in serum, 1:16 and 1:4, respectively. Trough sera contained little or no measurable antibiotic and had no antibacterial activity. Both cephalothin and ceftazidime were efficacious in the treatment of infected rabbits. There were no statistically significant differences in efficacy as defined by survival, eradication of bacteremia, or sterilization of cardiac vegetations. Results of laboratory tests which quantitated antimicrobial activity did not correlate with efficacy, either independent of antibiotic or adjusted for antibiotic. Despite their lesser in vitro activities, the new cephalosporins may be equivalent to the older cephalosporins for treating staphylococcal infections in humans, when administered in maximum recommended doses.