We review here various ways in which cardiac assistance might be derived from a patient's own skeletal muscle. Calculations based on experimental data and optimistic estimates of the efficiency of the energy conversions involved suggest that the continuous assist available would be limited to about 2 litres a minute if a muscle were used to energise an electromechanical device. It would be more efficient to couple the energy mechanically or hydraulically, but these approaches still pose problems of anatomical placement, muscle attachment, fluid leakage, and cost. Unless these issues can be addressed, the use of skeletal muscle as an internal power source for mechanical circulatory assist devices will remain an unworkable concept. Configurations that couple skeletal muscle contraction directly to the circulation would be more efficient and less costly. In terms of the energy available, a skeletal muscle ventricle could be designed to provide a continuous partial assist of 1–21/min, with flows of up to 8 1/min sustainable for limited periods. Such an approach offers new possibilities for the surgical treatment of chronic cardiac failure.