The recommended treatment of Lyme disease is evolving and important questions remain unanswered, such as (a) Are inexpensive oral regimens effective in curing acute illness and preventing arthritic, neurologic, and cardiac manifestations or are much more costly, and potentially toxic, intravenous antibiotics required? (b) Are relatively short 2- to 3-week courses of antibiotics sufficient or are prolonged regimens of a month, or more, better? This study reviews antibiotic therapy prescribed by Maryland physicians for the 283 cases reported in 1991 that meet the Centers for Disease Control and Prevention's case definition for Lyme disease. The purpose of the review was to obtain baseline information on the antibiotics being used by physicians in practice to treat patients that they believe have Lyme disease. The most frequently prescribed antibiotics for either the 60 percent of patients presenting with erythema migrans or the 40 percent with arthritic, neurologic, or cardiac manifestations were oral doxycycline (47 percent), tetracycline (11 percent), and amoxicillin (13 percent). Seventy-one percent of therapeutic courses were for 2 to 3 weeks. Amoxicillin was used in two-thirds of children younger than 8 years. Sixty (21 percent) received intravenous therapy, of which ceftriaxone, with or without other antibiotics, was almost always (95 percent) used. Intravenous therapy was more frequently given to those with arthritic, neurologic, and cardiac manifestations than to those with erythema migrans (odds ratio = 3.7) and to those with these systemic symptoms along with erythema migrans than to those with erythema migrans alone (odds ratio = 3.8). The average course was 2 days longer in treating those with arthritic, neurologic, or cardiac manifestations than in treating those with erythema migrans alone(P = 0.05).An epidemiologic assessment of antibiotics prescribed by the physicians in Maryland to treat Lyme disease in 1991 shows the choices, dosage, and duration of drugs generally followed those most frequently recommended in the literature. Also, it shows that efforts to educate physicians should be directed more towards the diagnosis rather than the treatment of Lyme disease.