Background: The value of symptoms and signs in the diagnosis of CHF has rarely been tested in large numbers of patients in the community. The aim of this study was to evaluate the importance of symptoms, signs, and past medical history in the diagnosis of CHF in primary care. Methods: Data on a sample of Portuguese men and women attending 365 primary care centres for any condition other than the treatment of acute infection, metabolic conditions or pregnancy were collected. All subjects who scored three or more points in the sum of categories one and two of the Boston questionnaire (history and physical examination) and those being treated for heart failure with loop or thiazide diuretics were considered to have possible heart failure and referred for further assessment including a resting echocardiogram. The sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratio (LR) for the diagnosis of heart failure were calculated. Results: A total of 5434 subjects were identified, of whom 1058 fulfilled the criteria for further assessment; 551 subjects had cardiac dysfunction at rest, of which 35.5% were in NYHA class I and 4.9% in class IV. Prior use of digoxin (LR 24.9) and/or diuretics (LR 10.6), a history of coronary artery disease (LR 7.1) or of pulmonary oedema (LR 54.2), were associated with a greater likelihood of having heart failure. Amongst current symptoms, a history of paroxysmal nocturnal dyspnoea (LR 35.5), orthopnea (LR 39.1) and breathlessness when walking on the flat (LR 25.8) were associated with a diagnosis of heart failure. However, these symptoms were not frequent amongst patients with heart failure within this population (sensitivity b36%). Jugular pressure N6 cm with hepatic enlargement, and oedema of the lower limbs (LR 130.3), a ventricular gallop (LR 30.0), a heart rate above 110 bpm (LR 26.7), and rales at pulmonary auscultation (LR 23.3) were all associated with a diagnosis of heart failure, but were infrequent findings in patients with heart failure (sensitivity b10%). Conclusions: Symptoms and signs, and clinical history had limited value in diagnosing heart failure when used alone. The signs and symptoms that best predicted a diagnosis of heart failure were those associated with more severe disease. If investigation is limited to patients with more definite symptoms and signs of heart failure, fewer than 50% of cases will be identified and a large number of patients with mild symptoms will be missed.