The diagnosis of vascular dementia (VaD) remains a controversial issue in many aspects and concepts. These nosologic problems are caused both by the methods, insufficient to ascertain the diagnosis, as well as by the weak consistency of the clinical concept of VaD itself. One of the most intriguing issues on VaD, and in particular on post-stroke dementia (PSD), is related to the time of development of cognitive decline. In clinical practice, the 3-month time threshold is usually chosen to enable resolution of a possible acute post-stroke delirium, and to obtain a more reliable cognitive assessment with a complete regression of acute neuropsychological stroke-related deficits. Another relevant issue is the possibility to predict which patient will develop PSD. In this regard, recent data indicate an overlap between Alzheimer's disease (AD) and PSD, which seems to share risk factors and neuropathology. In most population samples these two disorders appear together, and the consensus is growing that a degenerative component has a more important role in determining PSD onset shortly after stroke than previously recognized. Therefore, anamnestic data have a fundamental role in this prognostic approach.