This study characterized the relationship between clinical response, serum rituximab concentrations, and peripheral B-cell levels in patients with rheumatoid arthritis treated with rituximab. Data were analyzed from a double-blind, phase IIa trial in which 161 patients with active rheumatoid arthritis despite continuing methotrexate were randomized to methotrexate alone (10-25 mg/wk), rituximab alone (single course: 1000 mg administered intravenously on days 1 and 15), rituximab plus cyclophosphamide (750 mg administered intravenously on days 3 and 17), or rituximab plus methotrexate. Serum samples for pharmacokinetic analysis were collected through 24 weeks, and peripheral circulating CD19+ B-cell levels were measured through 48 weeks. All treatments were generally well tolerated, with no clinically relevant excess of adverse events leading to withdrawal among patients who received rituximab compared with those who received methotrexate alone. The proportions of patients who achieved an American College of Rheumatology score of 50 at week 24 were 13% (methotrexate alone), 33% (rituximab alone), 41% (rituximab plus cyclophosphamide), and 43% (rituximab plus methotrexate). Peripheral B-cell depletion occurred by day 15 in all patients treated with rituximab. There was no relationship between B-cell depletion and clinical response. Recovery of peripheral B cells was variable and showed no relationship with return of disease activity in patients who responded to rituximab. The mean terminal half-life of rituximab was 19 to 22 days; pharmacokinetic parameters were similar whether rituximab was administered alone or with methotrexate or cyclophosphamide. Because the level of peripherally circulating B cells does not appear to correlate with a maintained clinical response in patients with rheumatoid arthritis, the timing of rituximab retreatment should be based on clinical symptoms rather than peripheral B-cell levels.