Although syncope is common in the elderly, little is known of its epidemiology and prognosis. A retrospective analysis of syncope in 711 very old (mean age 87 years) institutionalised patients revealed a 10 year prevalence of 23 per cent and one year incidence of 7 per cent. A two-year prospective follow-up of this population revealed a yearly incidence of 6 per cent and recurrence rate of 30 per cent. Of 67 patients who developed syncope during follow-up, a cause was established in 46; 14 (21 per cent) had cardiac and 32 (48 per cent) had non-cardiac aetiologies. Twenty-one cases (31 per cent) remained unexplained. Patients who developed syncope were initially more functionally disabled (p = 0.003) and subsequently changed function more frequently (p = 0.03) than those without the development of syncope, but two year rates of hospitalisation and death were not different between the two groups. Life-table survival analysis showed no difference in the mortality of subgroups with cardiac, non-cardiac, and unknown aetiologies of syncope. Syncope is common in multiply impaired elders and is likely a manifestation of co-morbid disease rather than an independent contributor to mortality.