These studies extend the recent observation that cefazolin is inactivated to a greater extent than cephaloridine by some strains of penicillinase-producing Staphylococcus aureus, whereas cephalothin undergoes little if any inactivation. In Mueller-Hinton broth (inoculum, 3 x 10(6)) 100 recently isolated strains had minimal inhibitory concentrations (MICs) </= 2 mug/ml for cephalothin and cephaloridine, whereas in Trypticase soy broth (TSB) 50% had MICs > 2 mug/ml and 10% (designated "resistant" strains) were >8 mug/ml for cephaloridine but remained </=2 mug/ml for cephalothin. A large inoculum (3 x 10(7)) of strains with high MICs in TSB almost completely inactivated 50 mug of cefazolin per ml in 6 h, with progressively less inactivation, in the following order, of cephaloridine, cephalexin, cephradine, cephapirin, and cefamandole; cefoxitin and cephalothin underwent little if any inactivation. The greater inactivation in TSB than in Mueller-Hinton broth appeared to be due to a greater production of beta-lactamases by each colony-forming unit, since the inoculum size in the two broths was not significantly different. In contrast, "susceptible" strains (MICs </= 2 mug/ml in both broths) inactivated cephaloridine more than cefazolin, and equal amounts of powdered bacterial extracts confirmed the fact that qualitatively different beta-lactamases were produced by the susceptible and resistant strains. Disk diffusion tests were unreliable in separating the two groups of staphylococci. The clinical significance of inactivation by strains with high MICs is not known but, unless susceptibility can be clearly established, cephalothin appears preferable for severe staphylococcal infections, since it undergoes little if any inactivation by any strains of staphylococci.