Plain Language Summary What is the context? Relative thrombopoietin deficiency is implicated in primary immune thrombocytopenia (ITP), which is characterized by increased platelet destruction and impaired megakaryopoiesis. Patients who are innately unresponsive to or have relapsed after glucocorticoid treatment have limited treatment options. Recombinant human thrombopoietin (rhTPO) improves treatment response of primary ITP patients when added to high-dose dexamethasone. What is new? This trial sought to identify an optimal dosing regimen of rhTPO for patients who had failed or relapsed after glucocorticoid therapy. Of the 4 regimens, once daily 15000 U rhTPO for 14 injections yielded the greatest median increase in platelet count (167.5 x 10(9)/L) from baseline and attained the highest total response rate on day 14 (63.2%). 30000 U rhTPO once every other day for 7 injections was effective in rapidly increasing platelet counts in the first 7 days. All 4 regimens were safe and well-tolerated. What is the impact? The 30000 U rhTPO once every other day regimen may offer an effective and safe regimen with less frequent injections, but future trials with longer follow-up are needed. Recombinant human TPO (rhTPO) is effective for refractory/relapsed primary immune thrombocytopenia (ITP), but optimal dosing regimen remains elusive. In this multicenter, randomized, controlled trial, a total of 282 adult ITP patients (mean age 47.3 years; 82 men) with a platelet count <= 30 x 10(9)/L or >30 x 10(9)/L with active bleeding randomly received a once daily (QD) subcutaneous injection of 7500 U (n = 64) or 15000 U rhTPO for 14 injections, or 15000 U or 30000 U rhTPO once every other day (QOD) for 7 injections. The primary outcomes included change from baseline in platelet count and total response rate (TRR) on day 14. On day 14, the median increase of platelet count from baseline was the highest in the 15000-U QD group (167.5 x 10(9)/L, interquartile range [IQR] 23.0-295.0 x 10(9)/L), followed by the 30000-U QOD group (57.5 x 10(9)/L, IQR 9.0-190.0 x 10(9)/L) (ANCOVA P < .001; P = .266 with baseline count as a covariate). The TRR on day 14 was also the highest in the 15000-U QD group (63.2%), followed by the 30000-U QOD group (59.7%). The rate of grade 3 and above adverse events did not differ among the four groups. There were no new safety concerns. All 4 regimens are safe and well-tolerated. The 30000-U QOD regimen is practically indistinguishable in efficacy to the 15000-U QD regimen.