Background: Previous studies demonstrated that centralizing acute stroke treatment increases the chance of timely intra-venous treatment and lowers costs compared to care at community hospitals. This study aimed to estimate cost-effectiveness of centralizing acute stroke care in the Northern Netherlands using linked observational data.Methods: This study used observational data from 267 and 780 patients in a centralized and decentralized system, respectively. The original dataset was linked to the hospital information systems to estimate actual healthcare costs and EuroQol-5D questionnaire (EQ5D) utility values up to three months post stroke. EQ5D utility values were mapped from prospectively collected disability scores. Differences between the stroke care systems were estimated parametrically and using propensity score matching.Results: Mean healthcare costs up to three months were $US 7,790 (CI, 6,963 u2013 8,996) for the centralized system compared to $US 9,523 (CI, 8,923 u2013 10,123) for the decentralized system (P = 0.010). The mean EQ5D utility value at three months was 0.65 (CI, 0.63 u2013 0.67) for the decentralized system and 0.69 (CI, 0.65 u2013 0.73) for the centralized system (P = 0.064). The dominant effect remains stable after correcting for differences in systems and altering assumptions underlying cost derivation. Conclusions: In a real world setting a centralized system for acute stroke care appeared both cost-saving and yields better health outcomes, providing a strong rationale for centralization of acute stroke care systems.