The effect of dosage and the relative importance of peak and trough concentrations in serum for efficacy of teicoplanin were examined in a rabbit model of aortic valve endocarditis. Concentrations of teicoplanin in serum exceeded the MIC by several hundredfold, yet teicoplanin was less rapidly bactericidal than penicillin both in vitro and for endocarditis caused by a strain of Streptococcus sanguis. Because teicoplanin was 90% protein bound in rabbit serum, low free-drug concentrations probably resulted in less activity in vivo than in vitro. Because teicoplanin has a relatively low bactericidal rate and a high degree of protein binding, a sustained concentration in serum several times greater than the MIC may be important for efficacy in vivo. An intravenous regimen with relatively high peak concentrations in serum was less effective than an intramuscular regimen for endocarditis caused by a strain of Staphylococcus aureus, indicating that high peaks are unlikely to be an important determinant of efficacy. The therapeutically more relevant concentration in serum may be the trough.