Objective: The 5-level EuroQoL (EQ-5D-5L) is a patient-reported outcome measure frequently used in stroke research. However, it does not assess the cognitive problems many patients with stroke experience. The aim of this article is to compare the content validity, internal consistency and discriminative ability of the EQ-5D-5L with and without an additional cognitive domain (EQ-5D-5L+C), administered three months post-stroke. Design: Cross-sectional study. Setting: Six general hospitals in the Netherlands. Subjects: In all, 360 individuals with stroke three months after the event. Interventions: Not applicable. Main measures: The modified Rankin Scale and EQ-5D-5L+C were administered in telephone interviews three months post-stroke. Results: A total of 360 patients with stroke were included. Mean age was 68.8 years (standard deviation (SD) = 11.7), 143 (40%) were female, 334 (93%) had had an ischemic stroke, 165 (46%) had a National Institutes of Health Stroke Scale (NIHSS) score ⩽ 4 at presentation and the Barthel Index was 17.2 (SD = 4) four days post-stroke. Cognitive problems were reported by 199 (55%) patients three months post-stroke. Internal consistencies of the EQ-5D-5L and EQ-5D-5L+C were 0.75 and 0.77, respectively. Adding a cognitive domain resulted in a decrease of the ceiling effect from 22% to 14%. Both EQ-5D-5L and EQ-5D-5L+C showed good discriminative ability, but differences between patients with different modified Rankin Scale scores and with/without reported decrease in health and daily activities were slightly larger with the EQ-5D-5L+C compared to the EQ-5D-5L. Conclusions: The EQ-5D-5L+C, which includes a cognitive domain that is highly significant for stroke patients, showed increased content validity and good discriminative ability, without losing internal consistency.