We review current knowledge on the true size of the clinical condition known as 'heart failure' in terms of epidemiological information and in relation to the true clinical burden. Population studies, together with data from physician and general practitioner records, reveal a range of estimated heart-failure prevalence of 1-10%. Estimated incidence rates vary from approximately 0-1% per annum. Reasons for variation include age, sex and, possibly, methodology. In community studies, the five-year mortality is between 50-60% while, in patients requiring hospital admission, the annual mortality is 10-20% in those with mild-moderate symptoms, and as high as 40-60% in severe heart failure. While angiotensin-converting enzyme (ACE) inhibitor treatment does significantly improve mortality in all grades of symptomatic heart failure, the annual mortality in severe patients was still 36% in CONSENSUS I. It is obvious that the term 'heart failure' is insufficiently descriptive or specific to be an acceptable label for all patients who might benefit from treatment. As 'clinical' heart failure is often an advanced and irreversible state, studies of its antecedents are important in developing strategies aimed at retarding the progression from the asymptomatic to the symptomatic conditions. The true prevalence of the most common antecedent, left ventricular dysfunction, has received relatively little attention.