OBJECTIVE To assess energy depletion in skeletal muscle in patients with congestive heart failure by measuring blood purine metabolites during exercise and, at the same time, determine the implications of the ammonia response to exercise in these patients. METHODS Tottori University Hospital, Yonago, Japan. METHODS 49 heart failure patients (New York Heart Association (NYHA) grades I-III) and 16 normal subjects. METHODS Blood lactate, ammonia, and hypoxanthine levels were measured during exercise with expired gas analysis. RESULTS In normal exercising subjects as well as in each heart failure subgroup, the ammonia threshold was significantly higher than both the lactate threshold [control: 21.8 (SD 5.3) v 17.4 (3.3) ml/kg/min; NYHA class I: 18.9 (3.8) v 15.5 (2.6); class II: 14.8 (2.5) v 12.7 (2.4); class III: 13.5 (2.6) v 11.8 (2.5)] and the ventilatory threshold (P < 0.01). The difference between the ammonia and lactate thresholds was noted in all normal subjects and in all heart failure patients. The ammonia threshold, however, was significantly lower in heart failure patients than in normal subjects and it decreased with increasing NYHA class (P < 0.01). Maximum ammonia levels were lower in the heart failure group and decreased further with higher NYHA classifications [control: 198 (52) mg/dl; NYHA class I: 170 (74); class II: 134 (58); class III: 72 (15); P < 0.01]. There were significant correlations between maximum ammonia values and maximum lactate, oxygen consumption, and hypoxanthine levels (r = 0.74, 0.48, and 0.87, respectively; P < 0.001). CONCLUSIONS The ammonia threshold may reflect the onset of ATP depletion in exercising skeletal muscles, as opposed to the onset of anaerobic respiration. It seems therefore that energy depletion in skeletal muscles during exercise occurs after attaining the anaerobic threshold. Both aerobic and anaerobic capacities of skeletal muscle are reduced in patients with congestive heart failure.