OBJECTIVE To determine the efficacy of lysostaphin treatment of methicillin-sensitive and methicillin-resistant Staphylococcus aureus (MRSA) keratitis in a rabbit model. METHODS The sensitivity to lysostaphin and vancomycin were compared for 34 MRSA and 12 methicillin-sensitive strains. Methicillin-resistant S. aureus strain 301 (MRSA 301) or a methicillin-sensitive strain of low virulence, ISP546, was intrastromally injected into rabbit corneas. Rabbit eyes were treated topically every 30 minutes from 4 to 9 or 10 to 15 hours postinfection with 0.28% lysostaphin or 5.0% vancomycin. Rabbits were killed and corneas were excised and cultured to determine the number of colony forming units (CFU) per cornea. RESULTS Ninety percent minimal inhibitory concentrations were at least 19-fold lower for lysostaphin than for vancomycin. With early therapy (4 -9 hours postinfection) lysostaphin sterilized all MRSA 301-infected corneas, whereas untreated corneas contained 6.52 log CFU/cornea (P < or = 0.0001). Corneas infected with MRSA 301 and treated similarly with vancomycin retained 2.3 +/-0.85 log CFU/cornea, and none were sterile. When therapy was begun later (10-15 hours postinfection) the residual bacteria in lysostaphin-treated eyes were significantly less numerous than in vancomycin-treated eyes (0.58 +/- 0.34 vs. 5.83 +/- 0.16 log CFU/cornea, respectively; P < or = 0.0001). Three experiments were performed to demonstrate that lysostaphin penetrated the cornea to kill bacteria in vivo; lysostaphin-treated eyes were found to recover from infection, bacteria that did not cause epithelial defects (ISP546) were susceptible to lysostaphin, and inhibition of lysostaphin when harvesting corneas did not alter the observed therapeutic values of lysostaphin. CONCLUSIONS Lysostaphin is very effective in treating keratitis mediated by methicillin-sensitive or methicillin-resistant S. aureus.