This study compared co-amoxiclav, vancomycin and teicoplanin with and without netilmicin or amikacin for treating experimental subcutaneous fibrin-clot infection in rabbits due to a clinical beta-lactamase-positive methicillin- and gentamicin-resistant Staphylococcus epidermidis strain (MGRSE). MICs (mg/L) for this strain were: oxacillin 125, gentamicin 32, vancomycin 4, teicoplanin 8, netilmicin 1, amikacin 4, amoxycillin 64 with clavulanate at 2 mg/L. In rabbits treated with a single-dose i.v. regimen (netilmicin 8 mg/kg, amikacin 20 mg/kg, vancomycin 30 mg/kg, teicoplanin 15 mg/kg, co-amoxiclav 150-30 mg/kg), the bacterial count 24 h post-dose was reduced whatever the combination used (ANOVA, P < or = 0.001). Regimens were statistically classified in decreasing order of efficacy as follows: co-amoxiclav combined with netilmicin > vancomycin either alone or combined with either netilmicin or amikacin, teicoplanin with netilmicin > netilmicin and co-amoxiclav alone > teicoplanin or co-amoxiclav combined with amikacin, and teicoplanin alone > amikacin > no drug. From these findings, it is concluded that: co-amoxiclav could be useful for the treatment of beta-lactamase-positive and methicillin-resistant S. epidermidis infection; some enzyme-resistant aminoglycoside could be considered for treating gentamicin-resistant but netilmicin/amikacin-sensitive S. epidermidis infection; the combination of co-amoxiclav with netilmicin was synergistic and more rapidly bactericidal than vancomycin in this animal model.