During working contractions, chemical energy in the form of ATP is converted to external work. The efficiency of this conversion, called 'contraction coupling efficiency', is calculated by the ratio of work output to energy input from ATP splitting. Experiments on isolated muscles and permeabilized fibres show the efficiency of this conversion has a wide range, 0.2-0.7. We measured the work output in contractions of a single human hand muscle in vivo and of the ATP cost of that work to calculate the contraction coupling efficiency of the muscle. Five subjects performed six bouts of rapid voluntary contractions every 1.5 s for 42 s (28 contractions, each with time to peak force < 150 ms). The bouts encompassed a 7-fold range of workloads. The ATP cost during work was quantified by measuring the extent of chemical changes within the muscle from (31)P magnetic resonance spectra. Contraction coupling efficiency was determined as the slope of paired measurements of work output and ATP cost at the five graded work loads. The results show that 0.68 of the chemical energy available from ATP splitting was converted to external work output. A plausible mechanism to account for this high value is a substantially lower efficiency for mitochondrial ATP synthesis. The method described here can be used to analyse changes in the overall efficiency determined from oxygen consumption during exercise that can occur in disease or with age, and to test the hypothesis that such changes are due to reduced contraction coupling efficiency.