Weight change, subsequent survival time and cause of death are reported from the Dutch Longitudinal Study among the Elderly. Data consist of a national sample of persons aged 65-99 years. Six hundred and fifty-eight subjects were examined in the baseline years 1955-1957 and were re-examined 5 years later. Vital status and cause of death were ascertained for 604 of these subjects through 1983. Those subjects who experienced a decline in body mass index (BMI, kg/m(2)) during the period of observation, were likely to be in poorer health and have a shorter survival time than those subjects with stable weight, regardless of initial BMI. Weight gain was associated with shorter survival time only in the age group 65-74 and in those with heart disease. Weight loss, on the other hand, was most likely to result in decreased survival time among those ultimately dying of stroke, pneumonia/influenza or heart disease. As such, weight loss may be an indicator of the severity of disease. The noted associations remained, even when those surviving less than two years were omitted from the analyses. Thus, in longer survivors, weight loss may be associated with decreased vitality and decreased ability to survive once a disease becomes apparent.