OBJECTIVE Few studies have evaluated the roles of reducing disability after stroke in predicting survival. This study aimed to investigate the effects of improvement in the Barthel Index (BI) and other prognostic factors on survival in patients with first-time noncardioembolic ischemic stroke. METHODS BI effectiveness was defined as the improvement of BI between initial stroke (within 3 days) and 2 months after stroke. Cox regression analysis and Kaplan-Meier methods were used to evaluate the predictive roles of various prognostic factors. RESULTS A total of 111 patients were enrolled. Mean age at the time of stroke was 68 (+ or - 11.2) years. Median follow-up time was 77.4 months. Mean initial BI was 36.1 (+ or - 28.5) and mean BI effectiveness was 46.9 + or - 29.0. Overall, 55 deaths (49.5%) of the cohort were ascertained. The BI effectiveness had significant effects on long-time survival while initial BI was not a significant predictor. Higher BI effectiveness led to lower risk of mortality (hazard ratio = 0.44, 95% CI 0.24 - 0.80, p = 0.007). Elder age was correlated with poor survival (overall p = 0.006). Subjects in the eldest age group (> or = 70 years) showed a significant elevated risk for death (hazard ratio = 3.42, 95% CI 1.18 - 9.92). There was a trend indicating that the smaller the lesion size, the more favourable the prognosis (overall p = 0.057). CONCLUSIONS BI effectiveness in the first 2 months after first-time noncardioembolic stroke was more informative than initial disability status for predicting long-time mortality. It highlights the potential benefit in maximizing functional performance in patients with stroke.