|Background: After a stroke cognitive impairments are currently assessed with a paper-andpencil neuropsychological assessment. Virtual Reality (VR) may probably be more ecologically valid. It is not yet clear how stroke patients (often prone to overstimulation) tolerate and experience VR. Objective: The aim of this study is to compare the feasibility, user experience and preference of immersive (HMD) and non-immersive (desktop) user interfaces in a stroke population and to explore the effect of time post stroke. Methods: Fifty-nine stroke patients shopped in a virtual supermarket, once with each user interface. Feasibility measures (i.e., ability to complete the task, total duration, total number of correctly detected products) were derived. Afterwards patients filled in a questionnaire regarding their user experience with each user interface and a questionnaire regarding their preference. Results: There was no significant difference in number of patients completing the task, total duration and number of products found between the user interfaces. The feeling of presence, transportation, flow, negative effects and the overall experience of patients were higher with the immersive compared to the non-immersive interface. No significant interaction was found between time post stroke onset and the user interfaces, except for the number of products found. The majority of the sample did not prefer either interface. Conclusion: VR (immersive as well as non-immersive) can be used in clinical practice, at different times post stroke. The user experience is in general better with an immersive interface. Future research should therefore establish the ecological validity of (particularly immersive) VR in neuropsychological assessment.