The muscle glycogen content of the quadriceps femoris muscle was determined in 9 healthy subjects with the aid of the needle biopsy technique. The glycogen content could be varied in the individual subjects by instituting different diets after exhaustion of the glycogen store by hard exercise. Thus, the glycogen content after a fat ± protein (P) and a carbohydrate-rich (C) diet varied maximally from 0.6 g/100g muscle to 4.7 g. In all subjects, the glycogen content after the C diet was higher than the normal range for muscle glycogen, determined after the mixed (M) diet. After each diet period, the subjects worked on a bicycle ergometer at a work load corresponding to 75 per cent of their maximal O2 uptake, to complete exhaustion. The average work time was 59, 126 and 189 min after diets P, M and C, and a good correlation was noted between work time and the initial muscle glycogen content. The total carbohydrate utilization during the work periods (54–798 g) was well correlated to the decrease in glycogen content. It is therefore concluded that the glycogen content of the working muscle is a determinant for the capacity to perform long-term heavy exercise. Moreover, it has been shown that the glycogen content and, consequently, the long-term work capacity can be appreciably varied by instituting different diets after glycogen depletion.