Iwo decades of research have failed to generate consistent insight into the specificity of poststroke depression (PSD). This is, at least in part, caused by methodological difficulties. Differences in symptom profile between PSD and depression with no or another medical cause were described, but no specific and unequivocal clinical picture has been established so far. Prevalence rates of PSD varied largely between studies. In community based studies using standardised diagnostic instruments for depression, relatively low prevalence rates were reported compared to inpatient or rehabilitation studies. PSD occurs most frequently in the first few months after stroke, while a new incidence peak may occur 2-3 years after stroke. Two systematic reviews on the relation between lesion location and depression did not support the claim that left hemisphere lesions are a risk factor for PSD. A new concept of vascular depression has been proposed, which relates depression in the elderly to acute or chronic damage to the cerebral vascular system. Future efforts should aim at increasing the uniformity of study designs, assessment tools should be further improved for use in cognitively impaired patients and appropriate control groups should be defined to study the characteristic features of PSD.