We conducted a retrospective survey to assess prescription practice, response rates to rituximab, and the predictive indicators for a response to rituximab therapy in patients with primary or secondary immune thrombocytopenic purpura (ITP). Data were collected retrospectively from 40 consecutive patients with ITP attending our hospital: 29 (72.5%) had primary ITP and 11 (27.5%) had secondary ITP. Rituximab was given either as four weekly injections (375 mg/m(2)) or two injections of 1,000 mg given 2 weeks apart in 30 and 10 patients, respectively. The primary objective was to evaluate overall (OR) and complete response (CR) to rituximab therapy for ITP. OR was excellent and rapid and similar when secondary ITP was excluded from the analysis: OR was achieved in 28 (71.8%) patients and CR in 22 (56.4%). But, at >6 months, of the 28 responders, only 10 of 22 of the evaluable responses (45.5%) were sustained. In addition, except for the nonsignificant occurrence of antinuclear antibodies, no clinical or biological factors were predictive for OR or CR after the rituximab therapy. Twelve patients received a second course of rituximab. Overall, rituximab therapy achieved a response in two thirds of the patients, but the responders exhibited a high rate of early relapses, with no obvious difference according to the regimen of administration or rituximab dose.