An open, randomized, prospective study was carried out to compare the clinical efficacy and safety of phenoxymethylpenicillin with that of minocycline in the treatment of erythema migrans. Sixty patients (minocycline 30, penicillin 30) were enrolled in the study. The two groups of patients were statistically homogeneous regarding age and sex distribution. IgG and IgM antibodies against Borrelia burgdorferi were determined by ELISA before and after treatment and 1 year thereafter. Thirty-nine patients completed the study. All these patients (penicillin 21, minocycline 18) who received a 21-day course of treatment were free of clinical symptoms of late Lyme borreliosis after 1 year. Serum antibodies against B. burgdorferi could be detected before treatment in 6/21 patients treated with penicillin and 3/18 patients treated with minocycline. After 1 year 8/39 patients were seropositive without any evidence of ongoing disease. In the remaining 21 patients treatment could not be completed with the initial antibiotic due to side effects (penicillin 9/30, minocycline 12/30). One patient, who stopped penicillin treatment at day 14 and one patient who stopped minocycline at day 4, developed fatigue and memory impairment within the observation period. A 3-week course of treatment with penicillin or minocycline is equally effective in treating patients with erythema migrans and preventing late symptoms of Lyme borreliosis.