Usefulness of cardiopulmonary exercise testing in the diagnosis of heart failure was evaluated in a study of 59 patients with clinically suspected heart failure (22 men and 37 women) and 75 healthy control persons (34 men and 41 women), aged 45-74 years. Patients were classified according to certainty of the diagnosis by the Boston criteria: 27 patients were "unlikely" to have heart failure (group I), 19 had "possible" (group II) and 13 had "definite" heart failure (group III). Oxygen consumption at peak exercise and at the appearance of the ventilatory threshold was lower in group III than in the control group (15.6 +/- 1.5 versus 26.0 +/- 0.8 ml/min/kg, p less than 0.001 and 11.8 +/- 1.5 versus 18.0 +/- 0.4 ml/min/kg, p less than 0.001, respectively). However, the distribution of oxygen consumption values was wide in both the control and patient groups, and considerable overlapping of values between study groups was observed. Thus, cardiopulmonary exercise testing appeared to be of limited value in the diagnosis of heart failure.