The mean survival period in a series of 216 multiple sclerosis deaths, which formed part of a large prevalence sample observed in the Grampian region of Scotland, was 24.5 years, with an insignificant difference between females (25.7 years) and males (23.5 years). A third of the patients survived for over 30 years after onset. The age at death ranged between 25-80 years, with majority of the deaths occurring in the seventh decade (37%). On comparing life expectancy with the Scottish general population using life tables, only a slight reduction in the short-term (less than 10 years from onset) survival was noted in all age groups, with the exception of those with onset over the age of 50 years. The long-term life expectancy was however markedly reduced in all age groups compared with the controls. The survival period could be accurately predicted from the degree of disability at a point in time, and could be correlated with a number of clinical features, the most important of which was the age at onset. Eighty five per cent of those with onset of multiple sclerosis over the age of 50 years died within 20 years. Patients with a cerebellar disturbance at onset survived the shortest, and those with a brainstem lesion or retrobulbar neuritis the longest; those with a pyramidal dysfunction had an intermediate prognosis. Other parameters which could be correlated with the survival were: the timing and frequency of occurrence of psychiatric and urinary symptoms, interval between onset and first relapse and the course of the disease. As expected, most patients (89%) were significantly disabled (unable to walk) prior to death, only a minority, however, had become so within 10 years of the onset (10%). Sixty two per cent of the patients died of complications of multiple sclerosis. No unusual excess of any disease was noted amongst other causes. As expected, the majority of patients (55%) had bronchopneumonia as the terminal event, 11% had septicaemia, 15% had myocardial infarction and 4% had documented pulmonary embolism. This is the largest series of its kind where prognosis, judged by survival period, has been assessed amongst all multiple sclerosis patients derived from a prevalence sample and observed till death.