<i>Background:</i> Obesity is believed to increase the risks of ischemic stroke or coronary heart disease; however, regarding outcome after established vascular diseases, recent unexpected evidence has suggested that an increased body mass index (BMI) might have beneficial effects (obesity paradox). The aim of this study was to evaluate the independent association between BMI and long-term mortality after ischemic stroke. <i>Methods:</i> A total of 1,592 consecutive patients with ischemic stroke were prospectively included through our stroke cohort. In this study, the levels of BMI were classified based on guidelines for the Asian-Pacific population. Initial neurological severity was estimated by the National Institutes of Health Stroke Scale (NIHSS) score. Information on mortality was collected until the end of 2009, and the median follow-up period was 4 years. To examine the association between BMI and long-term mortality, we used Cox’s proportional regression analysis. <i>Results:</i> During follow-up, 23% of patients died. Linear regression analysis showed that the level of BMI was inversely related to initial neurological severity (p = 0.002). In the model of adjustment of age and gender using Cox’s proportional regression analysis, this inverse trend was also significant (reference, normal weight; hazard ratio of underweight, 2.45; overweight, 0.77; obesity, 0.60). However, after adjustment of all covariates, including initial neurological severity, only the harmful effect of underweight remained significant (2.79; 95% CI, 1.92–4.05); however, beneficial effects of overweight and obesity did not. <i>Conclusion:</i> Our study showed that an independent association between BMI and long-term mortality after ischemic stroke was found only in underweight patients. The obesity paradox phenomenon seems to be limited, and might not be interpreted as a direct causal relationship due to a strong association with initial neurological severity.