Objective: To investigate the effect of on-demand glucocorticoid strategy on the occurrence and outcome of porcine anti-lymphocyte globulin (p-ALG) -associated serum sickness in aplastic anemia (AA) . Methods: The data of AA patients who received in the Anemia Diagnosis and Treatment Center of Haematology Hospital, CAMS & PUMC from January 2019 to January 2022 were collected. Among them, 35 patients were enrolled in the on-demand group, with the glucocorticoid strategy adjusted based on the occurrence and severity of serum sickness; 105 patients were recruited in the usual group by matching the age and disease diagnosis according to 1∶3 ratio in patients who received a conventional glucocorticoid strategy in the same period. The incidences, clinical manifestations, treatment outcomes of serum sickness, and glucocorticoid dosage between the two groups were analyzed. Results: The incidences of serum sickness in the on-demand group and the usual group were 65.7% and 54.3% (P=0.237) , respectively. The median onset of serum sickness was the same [12 (9, 13) d vs the 12 (10, 13) d, P=0.552], and clinical symptoms and signs, primarily joint, and/or muscle pain, fever, and rash were similar. Severity grades were both dominated by Grades 1-2 (62.8% vs 51.4%) , with only a few Grade 3 (2.9% vs 2.9%) , and no Grades 4-5. No significant difference in the serum sickness distribution (P=0.530) . The median duration of serum sickness was the same [5 (3, 7) d vs 5 (3, 6) d, P=0.529], and all patients were completely cured after glucocorticoid therapy. In patients without serum sickness, the average dosage of prophylactic glucocorticoid per patient in the usual group was (469.48 ±193.57) mg (0 in the on-demand group) . When compared to the usual group, the average therapeutic glucocorticoid dosage per patient in the on-demand group was significantly lower [ (125.91±77.70) mg vs (653.90±285.56) mg, P<0.001]. Conclusions: In comparison to the usual glucocorticoid strategy, the on-demand treatment strategy could significantly reduce glucocorticoid dosage without increasing the incidence of serum sickness; in addition, the duration of serum sickness and the incidence of above Grade 2-serum sickness were similar.