Abstract: Objective To compare EQ-5D and SF-6D for measuring health utility of stroke patients in health economic evaluation studies. MethodsA prospective cohort study was conducted on 596 stroke patients in the West China Hospital of Sichuan University from 2010 to 2016. Data were collected at baseline through face to face interviews and at the follow-up stages through telephone interviews with a three-month interval. EQ-5D and SF-6D were used for measuring health utility scores of the participants. The consistency of the two instruments was assessed using Bland-Altman plot and Intraclass correlation coefficient （ICC）. Logistic regression models were established to identify predictors of health utility. ResultsThe participants had a mean utility score of 0.78 (95% confidence interval:0.76,0.80) in EQ-5D, compared with 0.74 (95% confidence interval: 0.73,0.76) in SF-6D, and a median (interquartile range) of 0.86 (0.68,1.00) in EQ-5D and 0.73 (0.62,0.86) in SF-6D. The 95% limits of agreement between the two instruments ranged from -0.28 to 0.35, with an ICC of 0.67 (95% confidence interval: 0.62, 0.71). EQ-5D had a higher ceiling effect. The health utility score of stroke patients changed there rapidly in acute phase (less than 3 months) but barely changed there after.Severity of stroke was a major predictor of health utility scores. Conclusion The two instruments generate inconsistent results in health utility. SF-6D is better for measuring health utility in patients with stroke in China.