To test the hypothesis that a decrease in intramuscular glycogen availability may stimulate heat shock protein expression, seven men depleted one leg of muscle glycogen the day before performing 4-5 h of exhaustive, two-legged knee extensor exercise at 40 % of leg peak power output. Subjects then rested for a further 3 h. Muscle biopsies were obtained from the depleted and control leg before, immediately after and 3 h into recovery from exercise. These samples were analysed for muscle glycogen, and HSP72 gene and protein expression. In addition, catheters were placed in one femoral artery and both femoral veins and blood was sampled from these catheters prior to exercise and at 1 h intervals during exercise and into recovery for the measurement of arterial-venous differences in serum HSP72. Plasma creatine kinase (CK) was also measured from arterial blood samples. Pre-exercise muscle glycogen content was 40 % lower in the depleted compared with the control leg and this difference was maintained throughout the experiment (P < 0.05; main treatment effect). Neither HSP72 gene nor protein expression was different pre-exercise. However, both HSP72 gene and protein increased (P < 0.05) post-exercise in the depleted leg, but not in the control leg. Exercise did not increase plasma CK concentrations and we were unable to detect HSP72 in the serum of any samples. These results demonstrate that while acute, concentric exercise is capable of increasing HSP72 in human skeletal muscle, it does so only when glycogen is reduced to relatively low levels. Hence, our data suggest that HSP72 protein expression is related to glycogen availability. In addition, because CK did not increase and we found no evidence of HSP72 in the venous effluent, our data suggest that skeletal muscle is impermeable to HSP72.