The objective of this study was to investigate the value of screening for cognitive functions at the start of an inpatient rehabilitation programme to predict the health status 1 and 3 years poststroke. In this longitudinal cohort study of stroke patients in inpatient rehabilitation data of 134 participants were analysed. Cognitive and clinical data were collected after admission to inpatient rehabilitation treatment. Data on health status were collected at 1 and 3 years post stroke. Cognitive functioning was assessed with the Cambridge Cognitive Examination (CAMCOG). Health status was assessed with the Stroke-Adapted Sickness Impact Profile. Statistical analyses included descriptive statistics, Pearson correlations and multiple regression analysis. One year poststroke, the CAMCOG dimensions of orientation (beta = -0.21), Perception (beta = -0.16) and Memory (beta = -0.16), as well as age and motor function were significant predictors of health status. Cognitive variables alone explained 2.5-10.3% of the variance. Three years poststroke, CAMCOG Perception (beta = -0.19 and -0.18) and Language (beta = -0.15), as well as age, type of stroke and motor function were significant predictors of health status, and the cognitive variables alone explained 1.7-10.9% of the variance. CAMCOG scores were significant predictors of long-term health status in patients with stroke, although the amount of explained variance was small.