During biosimilar drug development, conducting a clinical trial of biosimilar efficacy in patients may become necessary in the presence of residual uncertainty regarding the biosimilarity of the drugs. In the development of the biosimilar romiplostim GP40141, we aimed to use a model‐based in silico clinical trial (ISCT) approach to optimize the planned biosimilar efficacy trial in patients with immune thrombocytopenia. The population pharmacokinetic/pharmacodynamic model for healthy volunteers was modified and validated to describe platelet dynamics in patients with immune thrombocytopenia. ISCTs were then conducted using the modified model for various expected scenarios of biosimilar efficacy trials. Statistical analysis of the simulation results was subsequently used to confirm the appropriateness of the chosen design for evaluating the planned efficacy end points. Since the planned trial includes both patients naïve to therapy with thrombopoietin receptor agonists and nonnaïve patients, various expected ratios of naïve to nonnaïve patients (1:1, 1:2, 1:3) and the percentage of nonnaïve patients who previously received eltrombopag (0% or 30%) were assessed across 200 ISCTs performed for each scenario. The obtained estimates of empirical power for the equivalence test of platelet response/durable platelet response by the 10th/26th week between the test and reference groups were not less than 94%, regardless of the scenario. Differences in power between the 10‐ and 26‐week end points did not exceed 4%. The analysis of ISCT results allowed for an effective reduction of uncertainty in the biosimilar development of GP40141, demonstrating the appropriateness of using the 10‐week efficacy end point as the primary one.