Objective Antigen escape, particularly CD19 loss, limits the efficacy of CD19-targeted chimeric antigen receptor T-cell (CAR-T) therapy in children with relapsed/refractory B-cell acute lymphoblastic leukemia (R/R B-ALL). PAX5 gene variations have been shown to disrupt CD19 expression stability and increase relapse risk. This study aimed to compare the therapeutic efficacy of two CD19/CD22 dual-targeting CAR-T strategies, namely co-infusion versus bicistronic construct, in children with R/R B-ALL harboring PAX5 gene variations. Methods A retrospective analysis was conducted on the clinical data of R/R B-ALL children with PAX5 gene variation admitted to the hospital from November 5, 2019 to May 31, 2023. The patients were divided into a co-infusion group (Co group) and a bicistronic group (Bi group). All children received lymphodepletion conditioning followed by CAR-T cell infusion at a dose ranging from 2×106 to 10×106 cells/kg. The primary endpoints included the 30-day complete remission (CR) rate, minimal residual disease (MRD) negativity rate, relapse rate, survival outcomes, and treatment-related toxicity. The follow-up cutoff date was March 31, 2025. Results A total of 22 children with R/R B-ALL and PAX5 gene variations were included, including 18 cases (81.8%) of fusion variations, 2 cases (9.1%) of deletions, and 2 cases (9.1%) of missense variations. Six children (27.3%) had complex karyotypes (≥ 3 abnormalities). Among them, 8 patients received combined infusion of CAR-T treatment (Co group), and 14 received bicistronic CAR-T treatment (Bi group). The median age of the children at the time of treatment was 5.57 years (ranging from 1.31 to 14.85 years), and there were 13 boys and 9 girls. The CR rate of children in both the Co group and the Bi group was 100% 30 days after infusion, and the MRD was negative in both groups (MRD<0.01%). The median follow-up time was 26.5 months (range 5 to 66 months). Among the 22 patients, the 6-month, 12-month and 3-year overall survival (OS) rates were 95.45%, 90.91% and 63.64%, respectively. Meanwhile, the 6-month, 12-month and 3-year event-free survival (EFS) rates were 77.27%, 54.55% and 40.91%, respectively. The 6-month, 12-month, and 3-year OS rates in the Co group were 100%, 87.5%, and 50%, respectively, while those in the Bi group were 92.9%, 92.9%, and 71.4%, respectively. The 6-month, 12-month, and 3-year EFS rates in the Co group were 75%, 62.5%, and 37.5%, respectively, and those in the Bi group were 78.6%, 50%, and 42.9%, respectively. There were no statistically significant differences in OS and EFS rates between the two groups (POS=0.411, PEFS=0.826). A total of 4 children received allogeneic hematopoietic stem cell transplantation (allo-HSCT) after CAR-T treatment. The 6-month, 12-month and 3-year OS rates in the transplantation group were 100%, 100% and 50.0%, respectively, while those in the non-transplantation group were 94.4%, 88.9% and 66.7%, respectively. The 6-month, 12-month and 3-year EFS rates in the transplantation group were 75.0%, 75.0% and 50.0%, respectively, while those in the non-transplantation group were 77.8%, 50.0% and 38.9%, respectively. There were no statistically significant differences in OS and EFS rates between the two groups (POS=0.693, PEFS=0.553). Regarding toxicity, the incidence of grade ≥3 cytokine release syndrome (CRS) was 25.0% and 57.1% in the Co and Bi groups, respectively; the incidence of immune effector cell-associated neurotoxicity syndrome (ICANS) was 25.0% and 21.4%, respectively. All adverse events were manageable and no treatment-related death occurred. Conclusions Both co-infusion and bicistronic CD19/CD22 dual-targeting CAR-T strategies exhibit favorable short-term efficacy and comparable safety profiles in pediatric R/R B-ALL with PAX5 variations. However, the 3-year EFS rate remains at a relatively low level, suggesting that the long-term prognosis still needs further improvement.