Objective: To investigate the value of platelet count in predicting the efficacy of rituximab treatment in chronic primary immune thrombocytopenia (ITP). Methods: A retrospective study was conducted in 103 chronic ITP patients hospitalized in our medical center between January 2011 and December 2014. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) of platelet count in different time points were analyzed for the predictor of treatment response. Optimal cutoff values were established using ROC analysis. Results: A total of 103 patients were included in the study. There were 46 males and 57 females, with a median age of 30 (18-67) years. At day 1, 3 and 7 after the first dose of rituximab, there was no significant difference in platelet counts between the success group (PLT≥50×10(9)/L after treatment) and the failure group (PLT≤50×10(9)/L after treatment) (P>0.05). At day 14 after rituximab treatment (PTD 14), platelet counts became significantly different in the success and failure groups[41(8-384)×10(9)/L vs 23(0-106)×10(9)/L, P=0.003], and remained different thereafter, with increasing significance in the subsequent follow-ups. Patients were divided further using an optimal cut-off platelet count of 50×10(9)/L on PTD 14, PTD 30, and PTD 60, and PPV and NPV values were calculated for predicting eventual success and failure. Conclusion: Response can be predicted by obtaining platelet counts at 14, 30 and 60 days after rituximab treatment. The study proposed a protocol that guides patient monitoring and management planning.