Imipenem was very active in vitro against 36 Staphylococcus aureus isolates from cases of infective endocarditis; the MBC90 was 0.06 mg/l (four- to eight-fold more active than nafcillin). The in-vitro activity of imipenem against 22 Streptococcus faecalis isolates from proven endocarditis cases was similar to that of penicillin G (MBC90 = 8 mg/l). Imipenem was compared with nafcillin and with penicillin plus gentamicin in the therapy of experimental endocarditis induced in rabbits by Staph. aureus and Str. faecalis, respectively. The dosages were chosen to simulate closely serum antibiotic concentrations found in humans receiving standard parenteral regimens. Imipenem was more rapidly bactericidal than nafcillin in experimental staphylococcal endocarditis. The mean +/- S.D. Staph. aureus concentrations within aortic valve vegetations (log10 cfu/g) after 5 days of therapy were as follows: imipenem = 1.39 +/- 0.61 versus nafcillin 2.39 +/- 0.36 (P less than 0.02). Both the imipenem and nafcillin regimens resulted in 'sterile' vegetations in congruent to 50% of rabbits with experimental staphylococcal endocarditis after 5 days of therapy (P greater than 0.05). Imipenem was also equivalent to penicillin plus gentamicin in the therapy of experimental enterococcal endocarditis for 5 days, as assessed by the mean cfu/g vegetation and the percentage of vegetations rendered sterile. However, 7 days of therapy cured experimental enterococcal endocarditis in 72% of rabbits receiving penicillin plus gentamicin versus 20% for imipenem alone (P less than 0.05). Imipenem deserves further evaluation in the therapy of infective endocarditis, both in experimental animal models of infection and in humans. This agent may prove useful in the therapy of staphylococcal endocarditis in a variety of difficult clinical situations. Therapy of enterococcal endocarditis with imipenem alone is not advisable, pending further data.