A new semisynthetic macrolide roxithromycin was evaluated for its potential use in the treatment of Lyme borreliosis. Using a macro-dilution broth technique, Borrelia burgdorferi was shown to be susceptible to roxithromycin with a minimal bactericidal concentration (MBC) of 0.06-0.25 microgram/ml. A systemic B. burgdorferi infection was established in gerbils; a dosage of greater than or equal to 25 mg/kg/day roxithromycin for 10 days eliminated the infection. A single blind, randomized multicenter study was performed to evaluate the efficacy of roxithromycin 150 mg b.i.d. versus phenoxymethyl-penicillin 1 g b.i.d. for 10 days in patients with uncomplicated erythema migrans. The study was interrupted when 19 patients had enrolled because of five treatment failures. All 5 patients had received roxithromycin; three patients had persisting or recurrent erythema migrans, one developed a secondary erythema migrans-like lesion and severe arthralgia and one developed neuroborreliosis. B. burgdorferi was isolated from skin biopsies after roxithromycin therapy from two patients with persistent erythema migrans and both isolates were still highly susceptible to roxithromycin (MBC = 0.03 microgram/ml). No treatment failures were seen in 10 patients treated with phenoxymethyl-penicillin. Roxithromycin is thus not recommended for treatment of Lyme borreliosis.