Rituximab induces B-cell depletion; therefore, it has been used in the treatment of immune thrombocytopenic purpura (ITP). The aim of this retrospective study was to evaluate the effectiveness of rituximab in the treatment of 89 patients with chronic ITP refractory to several treatments. All the patients had platelet counts <30 x 10(9)/l. They had received a median of five (2-13) previous treatments, and 47 had undergone splenectomy. Rituximab was administered i.v. at 375 mg/m(2) in four weekly doses in 77 patients, and 12 patients received 1-6 doses. Forty-nine patients (55.1%) reached platelet counts >50 x 10(9)/l; 41 (46%) achieved a complete response (CR; platelets >100 x 10(9)/l), and eight (9%) obtained a partial response (platelets 50-100 x 10(9)/l). Overall, 31 patients (35%) maintained response, including 15 patients in whom splenectomy failed, with a median follow-up of 9 months (2-42), 12 for more than 1 year. The unique predictor of a maintained response was to reach a CR. Heavily treated patients (more than three different previous treatments, including any corticosteroids) and those with longer ITP duration (>10 years from diagnosis) had a worse response. Non-splenectomized patients had a better early response rate than those splenectomized. Rituximab was well tolerated, with two fever episodes following infusion and two reports of skin rash. Rituximab induced clinical responses in multi-treated refractory ITP patients with little toxicity and should be considered as an early therapeutic option in this setting, even as an alternative to splenectomy in selected patients.