OBJECTIVES: The aim of this study was to assess the  clinical and economic benefits of the stroke units  compared with conventional care. METHODS: A Markov model was developed. Four clinical conditions were defined :  first ischaemic event,  first cerebral haemorrhage,  recurrent ischaemic , recurrent haemorrhagic . Three lines of management were identified after hospitalisation: rehabilitation, home, and geriatric institutions.  Three levels of incapacity were identified using the Barthel index. The clinical benefit was measured as survival without loss of independence (Barthel 95-100). The relative risks of death or institutionalisation, or death or dependence after passage through a stroke unit were analysed in comparison with conventional hospital services from the data available in the Cochrane meta-analyse on stroke units. All of the cost calculations were made from the perspective of the health care system. Expenditures were discounted at a rate of 5%. RESULTS: The mean cost per patient at 5 years was €30,983 in a conventional care unit and €34,638.  in a stroke unit.After 5 years, the cumulative expenditure for the care of 120,000 patients in conventional care units was €3.710 million. Where care is generalised in a stroke unit, this figure increases to €4.160 million. In the 5 years following hospitalisation, patients cared for in stroke units survive for more trimesters without sequalae than those cared for conventionally (11.6 vs. 8.28). Comparison of the additional cost and added efficacy gave a ratio of €1359 per year of life gained without disability (Barthel 95-100). €513). CONCLUSIONS:  The cost-effectiveness ratio between for stroke units is much lower than the threshold for cost-effectiveness considered acceptable by the international scientific community (€53,400). This extremely favourable ratio could justify organisational changes and the establishment of stroke units in the hospital environment.