In a double-blind placebo-controlled trial carried out from 1980 to 1982, 20 patients with established Lyme arthritis were assigned treatment with 2.4 million U of intramuscular benzathine penicillin weekly for three weeks (total, 7.2 million U) and 20 patients received saline. Seven of the 20 penicillin-treated patients (35 per cent) had complete resolution of arthritis soon after the injections and have remained well during a mean follow-up period of 33 months. In contrast, all 20 patients given placebo continued to have attacks of arthritis (P less than 0.02). In 1983, of 20 patients treated with intravenous penicillin G, 20 million U a day for 10 days, 11 (55 per cent) had complete resolution of arthritis and have remained well since. As compared with nonresponders, penicillin-responsive patients in both studies were more likely to have previously received antibiotics for erythema chronicum migrans (P less than 0.02) and less likely to have been given intraarticular corticosteroids during or at the conclusion of parenteral therapy (P less than 0.1). The Lyme spirochete was not cultured from synovium or joint fluid. We conclude that established Lyme arthritis can often be treated successfully with parenteral penicillin. However, neither of the regimens that we tested is uniformly effective, and further experience will be needed to determine the optimal course of therapy.