Background Sovleplenib, a novel spleen tyrosine kinase (SYK) inhibitor, showed promising safety and activity in patients with primary immune thrombocytopenia in a phase 1b/2 trial. We aimed to evaluate the efficacy and safety of sovleplenib in patients with chronic primary immune thrombocytopenia. Methods This randomised, double-blind, placebo-controlled, phase 3 trial (ESLIM-01) was done in 34 clinical centres in China. Eligible patients, aged 18–75 years, had chronic primary immune thrombocytopenia, an Eastern Cooperative Oncology Group (ECOG) performance status of 0–1, and received one or more previous treatments. Patients were randomly assigned (2:1) to receive oral sovleplenib or placebo, 300 mg once daily, for 24 weeks. Randomisation was stratified by baseline platelet counts, previous splenectomy, and concomitant treatment for anti-immune thrombocytopenia at baseline. The primary endpoint was durable response rate (proportion of patients with a platelet count of ≥50 × 109/L on at least four of six scheduled visits between weeks 14 and 24, not affected by rescue treatment) assessed by intention-to-treat. The trial is registered with ClinicalTrials.gov, NCT05029635, and the extension, open-label phase is ongoing. Findings Between Sept 29, 2021, and Dec 31, 2022, 188 patients were randomly assigned to receive sovleplenib (n=126) or placebo (n=62). 124 (66%) were female, 64 (34%) were male, and all were of Asian ethnicity. Median previous lines of immune thrombocytopenia therapy were 4·0, and 134 (71%) of 188 patients had received previous thrombopoietin or thrombopoietin receptor agonist. The primary endpoint was met; durable response rate was 48% (61/126) with sovleplenib compared with zero with placebo (difference 48% [95% CI 40–57]; p<0·0001). The median time to response was 8 days with sovleplenib compared with 30 days with placebo. 125 (99%) of 126 patients in the sovleplenib group and 53 (85%) of 62 in the placebo group reported treatment-emergent adverse events (TEAEs), and most events were mild or moderate. Frequent TEAEs of grade 3 or higher for sovleplenib versus placebo were platelet count decreased (7% [9/126] vs 10% [6/62]), neutrophil count decreased (3% [4/126] vs 0% [0/62]), and hypertension (3% [4/126] vs 0% [0/62]). Incidences of serious TEAEs were 21% (26/126) in the sovleplenib group and 18% (11/62) in the placebo group. There were no deaths in the study. Interpretation Sovleplenib showed a clinically meaningful sustained platelet response in patients with chronic primary immune thrombocytopenia, with a tolerable safety profile and improvement in quality of life. Sovleplenib could be a potential treatment option for patients with immune thrombocytopenia who received one or more previous therapy. Funding HUTCHMED and Science and Technology Commission of Shanghai Municipality.