Eltrombopag is an oral thrombopoietin receptor agonist for the treatment of thrombocytopenia. We aimed to compare the response to once daily eltrombopag versus placebo in patients with chronic immune thrombocytopenia during a 6-month period. We undertook a phase 3, double-blind, placebo-controlled study in adults with previously treated immune thrombocytopenia of more than 6 months' duration who had baseline platelet counts lower than 30 000 per μL. Patients were randomly allocated (in a 2:1 ratio) treatment with local standard of care plus 50 mg eltrombopag or matching placebo once daily for 6 months. Randomisation was done centrally with a computer-generated randomisation schedule and was stratified by baseline platelet count (≤15 000 per μL), use of treatment for immune thrombocytopenia, and splenectomy status. Patients, investigators, and those assessing data were masked to allocation. Dose modifications were made on the basis of platelet response. Patients were assessed for response to treatment (defined as a platelet count of 50 000-400 000 per μL) weekly during the first 6 weeks and at least once every 4 weeks thereafter; the primary endpoint was the odds of response to eltrombopag versus placebo. Analysis was by intention to treat. This study is registered at ClinicalTrials.gov, number NCT00370331. Between Nov 22, 2006, and July 31, 2007, 197 patients were randomly allocated to treatment groups and were included in the intention-to-treat analysis (135 eltrombopag, 62 placebo). 106 (79) patients in the eltrombopag group responded to treatment at least once during the study, compared with 17 (28) patients in the placebo group. The odds of responding were greater in patients in the eltrombopag group compared with those in the placebo group throughout the 6-month treatment period (odds ratio 8.2, 99 CI 3.59-18.73; p<0.0001). 37 (59) patients receiving eltrombopag reduced concomitant treatment versus ten (32) patients receiving placebo (p=0.016). 24 (18) patients receiving eltrombopag needed rescue treatment compared with 25 (40) patients receiving placebo (p=0.001). Three (2) patients receiving eltrombopag had thromboembolic events compared with none in patients on placebo. Nine (7) eltrombopag-treated patients and two (3) in the placebo group had mild increases in alanine aminotransferase concentration, and five (4) eltrombopag-treated patients (vs none allocated to placebo) had increases in total bilirubin. Four (7) patients taking placebo had serious bleeding events, compared with one (<1) patient treated with eltrombopag. Eltrombopag is effective for management of chronic immune thrombocytopenia, and could be particularly beneficial for patients who have not responded to splenectomy or previous treatment. These benefits should be balanced with the potential risks associated with eltrombopag treatment. GlaxoSmithKline.