The aim of the present study was to test the hypothesis that consuming protein does not attenuate AMPK signalling when exercise is commenced in a glycogen-depleted state. After performing a glycogen-depleting protocol the evening before, the subsequent morning ten active men performed 45 min steady-state cycling at 50 % of peak power output (PPO) followed by an exercise capacity test (1-min intervals at 80 % PPO interspersed with 1-min periods at 40 % PPO). In a repeated measures design, subjects consumed 20 g of a casein hydrolysate solution (PRO) 45 min before exercise, 10 g during and a further 20 g immediately post-exercise, or an equivalent volume of a non-calorie taste matched placebo (PLA). Resting (PRO = 134 ± 29; PLA = 136 ± 28 mmol kg(-1)) and post-exercise muscle glycogen (PRO = 43 ± 16; PLA = 47 ± 18 mmol kg(-1)) was not different (P > 0.05) between trials nor was exercise capacity (PRO = 26 ± 9; PLA = 25 ± 10 min, P > 0.05). Phosphorylation of AMPK(Thr172) increased threefold immediately post-exercise (P < 0.05) and PGC1-mRNA increased sixfold at 3 h post-exercise (P < 0.05), though there were no differences between conditions (P > 0.05). In contrast, there was a trend (P = 0.08) for a divergent response in eEF2(Thr56) phosphorylation such that 1.5 fold increases post- and 3 h post-exercise in PLA were blunted with PRO, thus indicative of greater eEF2 activation. We conclude that athletes who deliberately incorporate training phases with reduced muscle glycogen into their training programmes may consume protein before, during and after exercise without negating signalling through the AMPK cascade.