AbstractThe prognostic relevance of strokes in different locations is debated. For example, insular strokes have been associated with increased mortality, but this association could reflect their greater severity. In two independent cohorts of patients with supratentorial ischemic stroke (n = 90 and 105), we studied the prognostic consequences of lesion size and location using voxel-based lesion-symptom mapping before and after volume control, which better accounts for total lesion volume. Strokes affecting the insula were larger than non-insular strokes (28 vs 2cc and 25 vs 3cc, p < 0.001 in both cohorts). A number of supratentorial areas (mainly in the left hemisphere), including the insula, were associated with poor functional outcome in both cohorts before (4014 voxels) and after volume control (1378 voxels), while the associations with death were greatly reduced after volume control (from 8716 to 325 voxels). Exploratory analyses suggested that the method of lesion volume quantification, the National Institutes of Health Stroke Scale hemispheric bias and patient selection can result in false associations between specific brain lesions and outcomes. In conclusion, death in the first months after stroke is mainly explained by large infarct volumes, whereas lesions of specific supratentorial structures, mostly in the left hemisphere, also contribute to poor functional outcomes.