Objective To assess the appropriateness of Barthel Index (BI) and Modified Rankin Scales (MRS) used as long-term outcome measures in a stroke data register and to investigate the correlation between cutoff points of the two scales in different stroke patients with and without disability. Methods Nine hundred and twelve patients were registered prospectively. BI and MRS were evaluated at the end of 1, 3, 6 and 12 months after stroke onset. The distribution, ceiling effects and floor effects of the two scales were evaluated. A logistic regression model was established to investigate correlation of cutoff points of BI and MRS. Results There were a total of 2 829 evaluation points of BI and MRS. The percentages of patients reaching the maximum scores of BI at the end of 3, 6 and 12 months (54.8％, 62.2％ and 68.3％, respectively) were higher than those of MRS. There was significant correlation between the two scales (Spearman’s correlation coefficient 0.887, P＜0.05), when MRS scores of ≤1 and ≤2 were taken as cutoff points, the corresponding cutoff points of BI score were ≥90 and ≥85, respectively. Conclusions BI has significant ceiling effects when used as long-term outcome measurement in a stroke data register. There was significant correlation between BI and MRS scores. In future clinical studies, an MRS score ≤2 or BI score ≥85 could be used as cutoff points in predicting stroke patients with and without disability.