Background: The 12-item Short-Form Health Survey (SF-12) was developed to measure perceived physical and mental health. Some studies of the psychometric properties, using classical test theory, of the SF-12 provide support for its use in patients with stroke, but it has not been scrutinized using recommended modern test theory approaches such as the Rasch measurement model among stroke survivors. Objectives: This study sought to explore the measurement properties of the SF-12 physical and mental health scales among people with stroke using the Rasch measurement model. Design: A cross-sectional design was used in this study. Methods: All patients discharged from a dedicated stroke unit in southern Sweden during 6 months were asked to participate 6 months later. Of 120 stroke survivors, 89 (74%) agreed to participate. Rasch analysis was used to assess the measurement properties of the SF-12 physical and mental component summary scores (PCS-12 and MCS-12, respectively). Results: For the PCS-12, we identified problems with targeting, overall and item-level fit, representing local response dependency, and multidimensionality. For the MCS-12, there were problems related to targeting (the persons felt better than the scale could conceptualize) and response categories that did not function as expected. However, MCS-12 items displayed reasonable model fit without indications of multidimensionality but with signs of local response dependency. Conclusion: The measurement properties of the MCS-12 in stroke appear reasonable unless milder mental health problems are of interest, whereas those of the PCS-12 are less acceptable. Given the interdependence between MCS-12 and PCS-12 that is inherent with the standard SF-12 scoring algorithm, such data should be interpreted with caution.