Using two strains of Staphylococcus aureus, one susceptible and one heterogeneously resistant to methicillin, for which MICs and MBCs of trimethoprim-sulfamethoxazole (TMP-SMX) were 0.06 and 0.06 micrograms/ml and 0.06 and 0.25 microgram/ml, respectively (concentrations are those of TMP), we studied the efficacies of TMP-SMX and cloxacillin, teicoplanin, and vancomycin for treatment of experimental staphylococcal endocarditis. Rabbits were treated with dosages of TMP-SMX selected to achieve concentrations in serum equivalent to that obtained in humans treated for Pneumocystis carinii pneumonia. The overall mortality rate of rabbits treated with TMP-SMX was 84% at day 3, not different from that of the control groups (P > 0.1). No sterile vegetations were observed to be present in control groups or in animals treated with TMP-SMX. However, 26, 60, and 75% of rabbits treated with teicoplanin, cloxacillin, and vancomycin, respectively, showed sterile vegetations. For methicillin-susceptible S. aureus (MSSA), the mean vegetation counts were not significantly different between the control group and the group treated with TMP-SMX (P > 0.1). For methicillin-resistant S. aureus (MRSA), treatment with TMP-SMX was more effective than no therapy, decreasing the number of organisms in vegetations (P < 0.01). For both strains, therapy with cloxacillin and therapy with teicoplanin or vancomycin were significantly more effective than therapy with TMP-SMX. Despite high concentrations of teicoplanin in serum which exceeded MBCs for staphylococci more than 50 times at the peak and 10 times at the trough, therapy with cloxacillin or vancomycin was superior to therapy with teicoplanin against both MSSA and MRSA. These data do not support the use of TMP-SMX in treatment of endocarditis and other severe staphylococcal infections with high bacterial counts.