To determine the prognosis, cause of death, and its determinants in participants of the population-based Rotterdam Study who were found to have heart failure. In 5255 Rotterdam Study participants (aged 68.9+/-8.6 years, 3113 women) the presence of heart failure was determined. Data were analysed with Cox's proportional-hazards models. One hundred and eighty-one participants (age 77.3+/-7.9 years, 109 women) had heart failure. Of these 85 (47%) died during the 4.8-8.5 (mean 6.1) years of follow-up. One, 2 and 5 years' survival was 89%, 79%, and 59%, representing an age-adjusted mortality twice that of persons without heart failure (hazard ratio 2.1, 95% CI 1.8-2.7). The hazard ratio for sudden death was even more pronounced: 4.8, (95% CI 2.6-8.7). Diabetes mellitus, impairment of renal function and atrial fibrillation were associated with a poor outcome. A higher blood pressure and body mass index conferred a more favourable prognosis in those with heart failure. Heart failure generally afflicts older subjects in the community, carries a poor prognosis, especially in the presence of concomitant diseases, and confers a fivefold increase in the risk of sudden death. Copyright 2000 The European Society of Cardiology.