INTRODUCTION: There are currently rapid developments in our ability to offer effective preventive and therapeutic interventions for sufferers of stroke. Concurrent pressure on National and local health care budgets demands that new interventions be evaluated for their cost-effectiveness before their widespread implementation. Clinical trials are of relatively short duration and therefore cannot directly measure the long-term impact of therapies. OBJECTIVE: To develop a flexible and comprehensive tool to assist rapid outcome evaluation of potential interventions in stroke care in four countries: France, Germany, UK, US. METHOD: Decision-analytic software was used to construct a Stroke Outcome Model, which consists of 4 inter-related modules: i. primary prevention, ii. ‘bridge’ to acute care, iii. acute care (ischaemic stroke only), iv. long-term care/prevention of recurrence in stroke survivors. Stroke prevention was represented by Markov state transition processes, states being defined in terms of use of therapy, disability and occurrence of (further) stroke. Data sources included clinical trials and meta-analyses, and prospective cohorts: Framingham study, Oxford Community Stroke Project and North Manhattan Stroke Study. Resource use data were obtained from published sources, supplemented by in-country panels of clinical experts, and unit costs from relevant National sources. RESULTS/DISCUSSION: The modelling approach has enabled the long-term consequences of acute or preventive therapy to be evaluated. Analyses have examined the cost-effectiveness of: antiplatelets in prevention of stroke recurrence; thrombolytic therapy in acute stroke; and stroke units. Results were sensitive to the duration of follow-up and the cost perspective chosen, in particular whether long-term care costs were included. The model enables users to explore issues such as the choice of effectiveness parameter or background event risks, and how risks are projected beyond the follow-up periods reported in trials or prospective studies. More data are required on long-term care, disability and quality of life/utility, and adjusted risks for (first) stroke.