Objective To compare the clinical efficacy， hematologic response kinetics， immunomodulatory effects， and safety profile of thrombopoietin receptor agonists （TPO-RAs） versus low-dose rituximab in adult patients with immune thrombocytopenia （ITP） refractory to or relapsing after first-line corticosteroid therapy. Methods This was a single-center， retrospective cohort study conducted at our institution between March 2021 and March 2025. A total of 80 adult patients diagnosed with primary ITP who exhibited inadequate response to ≥ 4 weeks of corticosteroid therapy or experienced relapse within 3 months after tapering were enrolled. Patients were stratified into two treatment cohorts based on clinical decision-making： the TPO-RA group （n = 55）， receiving either eltrombopag or romiplostim； And the low-dose rituximab group （n = 25）， receiving 100 mg weekly for four doses. Primary endpoints included overall response rate （ORR； defined as platelet count ≥ 30 × 10?/L and at least twofold increase from baseline） and time to response. Secondary endpoints encompassed platelet count trajectories， coagulation parameters?prothrombin time （PT） and activated partial thromboplastin time （APTT）?and immunophenotyping of peripheral blood T lymphocyte subsets： regulatory T cells （Tregs）， T helper 17 cells （Th17）， and the Treg/Th17 ratio. Adverse events were systematically recorded and graded per CTCAE v5.0. Results The TPO-RA group achieved a significantly higher overall response rate （85.5% vs. 64.0%， P = 0.032） and shorter median time to response （14 d vs. 28 d， P < 0.001） compared with the rituximab group. Post-treatment platelet counts were markedly elevated in the TPO-RA group ［（142.6 ± 48.3） × 10?/L vs. （98.1 ± 36.7） × 10?/L， P = 0.002］. No statistically significant differences were observed in PT （P = 0.417） or APTT （P = 0.352） between groups following therapy. Both regimens significantly increased circulating Treg frequency （P < 0.001 for both） and Treg/Th17 ratio （P < 0.001 for both）； however， the TPO-RA group demonstrated greater magnitude of improvement in both parameters relative to the rituximab group （Treg： 8.2% vs. 6.1%， P = 0.018； Treg/Th17： 2.4 vs. 1.7， P = 0.023）. Th17 cell frequencies did not differ significantly between groups post-treatment （P = 0.674）. The incidence of grade ≥ 2 adverse events was comparable （12.7% vs. 16.0%， P = 0.715）. Conclusions In corticosteroid-refractory or relapsed ITP， TPO-RAs confer superior and more rapid platelet recovery compared with low-dose rituximab， accompanied by more pronounced restoration of Treg-mediated immune homeostasis-without adversely affecting global coagulation function. These findings support TPO-RAs as a preferred second-line therapeutic option in this clinical setting.