Levofloxacin was investigated against viridans group streptococci in vitro and in rats with experimental aortic endocarditis. The MIC(90)s of levofloxacin and ciprofloxacin for 20 independent isolates of such bacteria were 1 and 8 mg/L, respectively. Rats were infected with two types of organism: either fully susceptible to levofloxacin MIC < or = 0.5 mg/L) or borderline susceptible (MIC 1-2 mg/L). Fully levofloxacin-susceptible bacteria comprised one penicillin-susceptible (MIC 0.004 mg/L) Streptococcus gordonii, and one penicillin-tolerant as well as one intermediate penicillin-resistant (MIC 0.125 mg/L) isogenic strains. Borderline levofloxacin-susceptible bacteria comprised one penicillin-susceptible Streptococcus sanguis and one highly penicillin-resistant Streptococcus mitis (MIC 2 mg/L). Rats were treated for 5 days with drug dosages simulating the following treatments in humans: (i) levofloxacin 500 mg orally once a day (q24 h), (ii) levofloxacin 500 mg orally twice a day (q12 h), (iii) levofloxacin 1 g orally q24 h, (iv) ciprofloxacin 750 mg orally q12 h, and (v) ceftriaxone 2 g iv q24 h. Levofloxacin was equivalent or superior to ceftriaxone, and was successful in treating experimental endocarditis irrespective of penicillin resistance. Nevertheless, standard levofloxacin treatment equivalent to 500 mg q24 h in human was less effective than twice daily 500 mg or once daily 1 g doses against borderline-susceptible organisms. Ciprofloxacin, used as a negative control, was ineffective and selected for resistant isolates. This underlines the importance of MIC determinations when treating severe streptococcal infection with quinolones. In the case of borderline-susceptible pathogens, total daily doses of 1 g of levofloxacin should be considered.