Purpose: State-based Medicaid programs have begun using All Patient RefinedﾖDiagnosis-Related Groups (APR-DRGs) to determine hospital reimbursement rates. Medicaid provides coverage for 45% of childhood cancer admissions. This study aimed to examine how well APR-DRGs reflect admission costs for childhood cancer chemotherapy to inform clinicians, hospitals, and policymakers in the wake of policy changes. Methods: We identified 25,613 chemotherapy admissions in the 2009 Kidsﾒ Inpatient Database. To determine how well APR-DRGs explain costs, we applied a hierarchic linear regression model of hospital costs, allowing for a variety of patient, hospital, and geographic confounders. Results: APR-DRGs proved to be the most important predictors of admission costs (P < .001), with costs increasing by DRG severity code. Diagnosis, age, and hospital characteristics also predicted costs above and beyond those explained by APR-DRGs. Compared with admissions for patients with acute lymphoblastic leukemia, costs of admissions for patients with acute myelomonocytic leukemia were 82% higher; non-Hodgkin lymphoma, 20% higher; Hodgkin lymphoma, 25% lower; and CNS tumors, 27% lower. Admissions for children who were 10 years of age or older cost 26% to 35% more than admissions for infants. Admissions to childrenﾒs hospitals cost 46% more than admissions to other hospital types. Conclusion: APR-DRGs developed for adults are applicable to childhood cancer chemotherapy but should be refined to account for cancer diagnosis and patient age. Possible policy and clinical management changes merit further study to address factors not captured by APR-DRGs.