To examine the ability of the skeletal muscle of congestive heart failure (CHF) patients to adapt to chronic exercise, five patients performed localized nondominant wrist flexor training for 28 d. Inorganic phosphate (Pi) and phosphocreatine (PCr) were monitored by magnetic resonance spectroscopy in both forearms at rest and during submaximal wrist flexion exercise at 6, 12, 24, and 36 J * min' before and after exercise training.Simultaneous measurements of limb blood flow were made by plethysmography at 12, 24, and 36 J * min'. Forearm muscle mass and endurance were measured by magnetic resonance imaging and wrist flexion exercise before and after training.The P1/PCr ratio and pH were calculated from the measured Pi and PCr. Exercise cardiac output, heart rate, plasma norepinephrine, and lactate measured during training were not elevated above resting values, confirming that training was localized to the forearm flexor muscles. After training, muscle bioenergetics, as assessed by the slope of the regression line relating P1/PCr to submaximal workloads, were improved in the trained forearm of each patient, although muscle mass, limb blood flow, and pH were unchanged. Forearm endurance increased by > 260% after training. In the dominant untrained forearm, none of the measured indices were affected. We conclude that localized forearm exercise training in CHF patients improves muscle energetics at submaximal workloads in the trained muscle, an effect which is independent of muscle mass, limb blood flow, or a central cardiovascular response during training. These findings indicate that peripheral muscle metabolic and functional abnormalities in CHF can be improved without altering cardiac performance. (J. Clin. Invest. 1990.