OBJECTIVE: An estimated 40% of patients experience cognitive impairment 3 months after a stroke. However, there is a lack of knowledge about cognitive improvement after stroke. This exploratory study aims to comprehend the prevalence and patterns of cognitive improvement within the initial 3 months poststroke. METHOD: At 1 week and 3 months poststroke, 85 patients aged 40-79 years with supratentorial ischemic stroke and a Mini-Mental State Examination score ≥23 underwent neuropsychological testing of learning, recall, recognition, working memory, verbal fluency, naming, executive function, attention, and processing speed. A hierarchical cluster analysis based on Ward's method and squared Euclidean distance was performed on changes in tests over 3 months to identify homogeneous groups. The within-cluster and between-cluster differences in change were presented descriptively for each cognitive domain. Potential predictors for cluster belonging were compared by independent samples t test or χ2 test. RESULTS: The cluster analysis identified two clusters. The improvement cluster (n = 22) progressed in learning, recall, recognition, executive function, and most measures of working memory, attention, and verbal fluency. In the stable cluster (n = 63), cognitive function remained stable across most domains, with an increase in a few working memory, verbal fluency and naming, and executive function measures. Notably, 67% of lacunar strokes and 93% of females were in the stable cluster, having less improvement. Age and education did not differentiate between the clusters. CONCLUSIONS: 26% of patients showed overall cognitive improvement within 3 months. Knowing the prevalence and patterns of cognitive improvement after stroke lays the groundwork for future research and clinical practice. (PsycInfo Database Record (c) 2025 APA, all rights reserved).