OBJECTIVES: to investigate patient and family carer preferences for stroke support services. METHODS: Discrete choice experiment (DCE) using a fractional factorial design, folded into nine binary choices, containing four attributes:  format of service (group, individual); service provider (hospital stroke team, primary care, voluntary sector); provision of additional social and leisure activities (provided, not provided); and time to plan and make the journey (1,2,4 hours).  Attributes and levels were defined by a review of policy documentation and two workshops with patients and family carers (n=8).  Patients (six months post-stroke) and their family carers were recruited from four stroke-services in Wales, to complete a paper questionnaire including the DCE and validated instruments to facilitate pre-specified sub-group analyses. Data were analysed in STATA using a random effects logit model.  Marginal rate of substitution (MRS) used journey time as the value attribute. RESULTS: 144/474 (30%) eligible patients requested the questionnaire, 80 (56%) completed (mean age 70.8 years (sd 11.1).  34/74 family carers who requested the questionnaire completed.  All four attributes were significant for patients (p<0.05), only format of service and journey for carers.  Patients preferred support services on an individual basis (p=0.00, MRS=128.61 minutes), with additional social and leisure activities (p=0.00, MRS=64.07 minutes).  Family carers were willing to trade more journey time for services provided on an individual basis (p=0.00, MRS=273.73 minutes).  Sub-group analysis showed significant differences between established (>315 days post-stroke) and early patients.  Models accounting for patient health utility were significantly different (p<0.05).    CONCLUSIONS: These findings complement the more descriptive summaries of a comprehensive matrix of sources of support within regions and localities (Department of Health, 2007), with which people living with stroke may interact. Exploring what patients and carers consider important in the provision of stroke support services, will ensure that service re-design is user-focused to maximum utility.