OBJECTIVES: Case mix reimbursement systems used in many countries to pay for hospital services typically rely on combinations of diagnoses and procedures to group patients according to their expected costs. In frail, complex populations such as patients with hip fracture, patients’ pre-hospitalization functional status and social circumstances may also be important predictors of resource utilization. We investigated the association between clinical, functional and social factors and hospitalization costs in order to inform the parameters for a new case mix payment system. METHODS: A multidisciplinary clinical expert panel identified candidate patient covariates to be modeled. We linked hospital, long-term care and home care records to examine patients’ pre-hospitalization functional status and place of residence. We employed generalized linear models to explore the association between patient characteristics and hospitalization costs. Based on the results of this analysis, the expert panel issued recommendations on the design of the new payment methodology.   RESULTS: The analysis cohort included 11,321 patients admitted with hip fracture to hospitals in Ontario, Canada throughout 2011/12. Of these, 17.6% were admitted from residence in long-term care. Multivariable analysis revealed the most important predictors of increased hospitalization costs to be impairments in activities of daily living, higher Charlson comorbidity score and pre-hospitalization residence in a long-term care home. Based on these results, the expert panel recommended a new hip fracture funding model that stratifies the population by pre-fracture place of residence and incorporates activities of daily living and comorbidity level as additional risk adjustors. CONCLUSIONS: While requiring record linkage across datasets, the inclusion of hip fracture patients’ pre-hospitalization characteristics as case mix adjustors can improve the performance of case mix payment systems. For the frail hip fracture population, functional and social factors should also be considered as predictors of hospitalization costs, alongside traditional clinically-focused variables.