Symptom-limited, laboratory-based exercise tests are often used to define prognosis in patients with chronic heart failure, but they do not relate to measures of normal daily activity. Invasive measures of central hemodynamics similarly relate poorly to outcome. Pedometer scores of weekly walking are markedly reduced in patients with heart failure, but whether this less artificial measure of exercise capacity is important in predicting prognosis is not known. Eighty-four patients with chronic heart failure were followed for a mean of 710 days during which 44 died and 3 underwent cardiac transplantation. Symptom-limited treadmill exercise capacity using 2 different protocols did not predict survival, whereas reduced weekly pedometer scores were strong predictors of death (p < 0.001). Other variables that predicted death included resting cardiac output, arterial blood pressure, diuretic requirements, New York Heart Association class, increased bilirubin, and hyponatremia (all p < 0.01). Reduced levels of daily activity are strong predictors of death in chronic heart failure and appear more powerful than laboratory-based exercise tests. This type of assessment is valuable in identifying patients at high risk and provides an objective measure of incapacity during normal daily life. The exercise capacity of patients unable to exercise in the laboratory could also be assessed using this technique. This may prove invaluable in clinical and mortality trials.