Validity of heart failure (HF) diagnosis was studied in 88 patients (37 men and 51 women), aged 45-74 (mean 61) years, in whom HF diagnosis had been newly made by primary health care physicians. Boston criteria for HF and a supplementary classification, based on information from clinical examinations and a 6-month follow-up, were used to define HF diagnosis as 'definite', 'possible' or 'unlikely'. Twenty-eight (32%) patients (21 men and seven women) had 'definite' HF and 46 (52%) (28 men and 18 women) had either 'definite' or 'possible' HF by both classifications. In 30 (34%) patients (six men and 24 women) HF diagnosis was 'unlikely' by both classifications. In conclusion, false-positive diagnosis of HF was common in primary health care, and HF diagnosis was more difficult in women than in men. Obesity, unrecognized symptomatic myocardial ischaemia without HF and pulmonary diseases were the most important conditions leading to false-positive HF diagnosis.