To determine whether chronic inflammatory arthritis may respond to antibiotic therapy (implying a bacterial origin), we conducted a placebo-controlled, double-blind study. Sixty patients with inflammatory arthritis and antibody titers to Borrelia burgdorferi 1:64 or more were randomized to receive placebo (n = 20) or 2 g/d of ceftriaxone intravenously (n = 40) for 2 weeks. Two of 20 placebo- and 19 of 40 antibiotic-treated patients improved. At 1 month, the placebo-treated patients could elect to receive ceftriaxone. Altogether, 58 patients were treated with ceftriaxone and followed up for 13 to 24 months. Improvement was noted in 27 of the 58 antibiotic-treated patients. Patients with a wide diversity of inflammatory arthritis were studied. Response to ceftriaxone was seen in all groups, including 5 of 12 with rheumatoid arthritis, 5 of 8 with psoriatic arthritis, 3 of 5 with vasculitis, and 14 of 33 with less well-differentiated chronic inflammatory arthritis. In 16 of the 27 who responded to the antibiotic, the arthritis worsened 6 to 18 months after the initial response to ceftriaxone. Previous improvement of arthritis after oral antibiotic was a better predictor of response to ceftriaxone than either duration of disease or Lyme antibody titer. Side effects to ceftriaxone were frequent and included diarrhea (29/60) and acute allergic reactions (9/58). We conclude that some patients may have an occult bacterial infection underlying their chronic inflammatory arthritis, and may respond to antibiotic therapy. The response to ceftriaxone in patients with even weakly reactive Lyme titers encourages further prospective placebo-controlled studies of antibiotics in various subsets of chronic arthritis.