Chronic immune thrombocytopenic purpura (ITP) is characterised by accelerated platelet destruction and decreased platelet production. Short-term administration of the thrombopoiesis-stimulating protein, romiplostim, has been shown to increase platelet counts in most patients with chronic ITP. We assessed the long-term administration of romiplostim in splenectomised and non-splenectomised patients with ITP. In two parallel trials, 63 splenectomised and 62 non-splenectomised patients with ITP and a mean of three platelet counts 30x10(9)/L or less were randomly assigned 2:1 to subcutaneous injections of romiplostim (n=42 in splenectomised study and n=41 in non-splenectomised study) or placebo (n=21 in both studies) every week for 24 weeks. Doses of study drug were adjusted to maintain platelet counts of 50x10(9)/L to 200x10(9)/L. The primary objectives were to assess the efficacy of romiplostim as measured by a durable platelet response (platelet count > or =50x10(9)/L during 6 or more of the last 8 weeks of treatment) and treatment safety. Analysis was per protocol. These studies are registered with ClinicalTrials.gov, numbers NCT00102323 and NCT00102336. A durable platelet response was achieved by 16 of 42 splenectomised patients given romplostim versus none of 21 given placebo (difference in proportion of patients responding 38% [95% CI 23.4-52.8], p=0.0013), and by 25 of 41 non-splenectomised patients given romplostim versus one of 21 given placebo (56% [38.7-73.7], p<0.0001). The overall platelet response rate (either durable or transient platelet response) was noted in 88% (36/41) of non-splenectomised and 79% (33/42) of splenectomised patients given romiplostim compared with 14% (three of 21) of non-splenectomised and no splenectomised patients given placebo (p<0.0001). Patients given romiplostim achieved platelet counts of 50x10(9)/L or more on a mean of 13.8 (SE 0.9) weeks (mean 12.3 [1.2] weeks in splenectomised group vs 15.2 [1.2] weeks in non-splenectomised group) compared with 0.8 (0.4) weeks for those given placebo (0.2 [0.1] weeks vs 1.3 [0.8] weeks). 87% (20/23) of patients given romiplostim (12/12 splenectomised and eight of 11 non-splenectomised patients) reduced or discontinued concurrent therapy compared with 38% (six of 16) of those given placebo (one of six splenectomised and five of ten non-splenectomised patients). Adverse events were much the same in patients given romiplostim and placebo. No antibodies against romiplostim or thrombopoietin were detected. Romiplostim was well tolerated, and increased and maintained platelet counts in splenectomised and non-splenectomised patients with ITP. Many patients were able to reduce or discontinue other ITP medications. Stimulation of platelet production by romiplostim may provide a new therapeutic option for patients with ITP.