OBJECTIVES: This research aimed to develop the risk adjusted cost-effectiveness (RAC-E) model that used routinely collected data to compare acute hospital services for stroke patients admitted to the main public hospitals. METHODS: Data sourced from routinely collected National Health Insurance Research claim Dataset (1997-2010). Patient cohort was defined by the ICD codes 430-438. Adverse events (AE) indicated by all-cause readmission or mortality after the index stroke. Univariate and multivariate associations between baseline characteristics and AE were assessed by chi-square tests and multiple Cox proportional hazards (Cox-PH) model. Survival curves adjusting for covariates were plotted for comparison of cumulative probability of AE. Cost-effectiveness analysis was performed by comparing the incremental cost-effectiveness ratios (ICER) which adjusted all the socioeconomic status and type of co-morbidities. RESULTS: With reference to hospital 0, hospital 1 was associated with a higher risk of incident adverse outcome, especially in intracerebral hemorrhage (ICH) (HR 1.63, p=0.01; OR 1.94, p<0.01) and cerebral infarction (CI) (HR 1.72, p<0.0001; OR 1.97, p<0.0001). After adjusting for age, gender and Charlson Comorbidity Index (CCI) scores, admission to Hospital 1 remained associating with increased AE risk in ICH (HR 1.48, p<0.05; OR 1.80, p<0.05) and CI (HR 1.61, p<0.001; OR 1.80, p<0.001). Our cost-effectiveness analysis demonstrated that Hospital 0 compared with Hospital 1 was considered to be a more cost-saving hospital at the incremental cost of 167,460 NTD for preventing one Subarachnoid hemorrhage (SAH), 19,947 NTD for preventing one ICH. Moreover, hospital 0 was more cost-effective than hospital 1 with an additional cost of 10,444 NTD for preventing one CI and 69,240 NTD for preventing one transient ischemic attack or other unspecified cerebrovascular disease (TIA). CONCLUSIONS: Overall, the present analysis indicated Hospital 0 performed better and may be a more cost-effective hospital for the care of stroke patients.