Key points: When muscle biopsies first began to be used routinely in research on exercise physiology five decades ago, it soon become clear that the muscle content of glycogen is an important determinant of exercise performance. Glycogen particles are stored in distinct pools within the muscles, but the role of each pool during exercise and how this is affected by diet is unknown. Here, the effects of diet and exercise on these pools, as well as their relation to endurance during prolonged cycling were examined. We demonstrate here that an improved endurance capacity with high carbohydrate loading is associated with a temporal shift in the utilisation of the distinct stores of glycogen pools and is closely linked to the content of the glycogen pool closest to actin and myosin (intramyofibrillar glycogen). These findings highlight the functional importance of distinguishing between different subcellular microcompartments of glycogen in individual muscle fibres. Abstract: In muscle cells, glycogen is stored in three distinct subcellular pools: between or within myofibrils (inter- and intramyofibrillar glycogen, respectively) or beneath the sarcolemma (subsarcolemmal glycogen) and these pools may well have different functions. Here, we investigated the effect of diet and exercise on the content of these distinct pools and their relation to endurance capacity in type 1 and 2 muscle fibres. Following consumption of three different diets (normal, mixed diet = MIX, high in carbohydrate = HIGH, or low in carbohydrate = LOW) for 72 h, 11 men cycled at 75% of (Formula presented.) max until exhaustion. The volumetric content of the glycogen pools in muscle biopsies obtained before, during, and after exercise were quantified by transmission electron micrographs. The mean (SD) time to exhaustion was 150 (30), 112 (22), and 69 (18) minutes in the HIGH, MIX and LOW trials, respectively (P < 0.001). As shown by multiple regression analyses, the intramyofibrillar glycogen content in type 1 fibres, particularly after 60 min of exercise, correlated most strongly with time to exhaustion. In the HIGH trial, intramyofibrillar glycogen was spared during the initial 60 min of exercise, which was associated with levels and utilisation of subsarcolemmal glycogen above normal. In all trials, utilisation of subsarcolemmal and intramyofibrillar glycogen was more pronounced than that of intermyofibrillar glycogen in relative terms. In conclusion, the muscle pool of intramyofibrillar glycogen appears to be the most important for endurance capacity in humans. In addition, a local abundance of subsarcolemmal glycogen reduces the utilisation of intramyofibrillar glycogen during exercise.