BACKGROUND Longitudinal studies on cognitive outcomes after stroke revealed heterogeneous results and the underlying pathology and risk factors for so-called post-stroke dementia are unclear. OBJECTIVE To assess long-term cognitive performance changes in patients after the first ischemic stroke and to evaluate possible risk factors for post-stroke dementia. METHODS In this study 66 clinically mildly affected patients aged 54-87 years without a history of dementia underwent extensive neuropsychological assessment after first ever ischemic stroke and again 6 months after the event (follow-up assessment). Demographic, clinical and paraclinical parameters were assessed as potential predictors for long-term cognitive outcome. RESULTS At the group level significant performance improvements were found for most of the neurocognitive domains at the follow-up assessment. The greatest cognitive improvement was found in visuospatial processing. Immediately after stroke 54.5% of patients were considered cognitively impaired (z-scores < -2 in at least 2 neurocognitive domains). At follow-up only 16.7% were considered cognitively impaired according to this criterion and among these only 2 patients (3%) had developed a new, clinically relevant cognitive impairment (i.e. post-stroke dementia). Patients with inferior cognitive performance improvements at follow-up had on average larger brain lesions caused by the stroke as well as a prediabetic metabolic status. CONCLUSIONS The probability of developing a post-stroke dementia syndrome is lower than previously assumed in patients with first ever stroke, with only mild clinical disability and without premorbid cognitive impairment. Long-term cognitive impairment could primarily be determined by the size of the lesioned brain area as well as the premorbid (pre)diabetic status.