Following fatigue, motor evoked potentials (MEP) elicited by transcranial magnetic stimulation (TMS) and cervicomedullary evoked potentials (CMEP) elicited by stimulation of the corticospinal tract are depressed. These reductions in corticomotor excitability and corticospinal transmission are accompanied by voluntary activation failure but this may not reflect a causal relationship. Our purpose was to determine whether a decline in central excitability contributes to central fatigue. We hypothesized that if central excitability limits voluntary activation, then a caffeine-induced increase in central excitability should offset voluntary activation failure. In this repeated-measures study, 8 males each attended 2 sessions. Baseline measures of knee extension torque, maximal voluntary activation, peripheral transmission, contractile properties, and central excitability were made prior to administration of caffeine (6 mg/kg) or placebo. The amplitude of vastus lateralis MEPs elicited during minimal muscle activation provided a measure of central excitability. Following a 1-hr rest, baseline measures were repeated before, during, and after a fatigue protocol which ended when maximal voluntary torque declined by 35% (T_lim). Increased prefatigue MEP amplitude (p=0.055) and cortically-evoked twitch (p<0.05) in the caffeine trial indicate that the drug increased central excitability. In the caffeine trial, increased MEP amplitude was correlated with time to task failure (r=0.74, p<0.05). Caffeine potentiated the MEP early in the fatigue protocol (p<0.05) and offset the 40% decline in placebo MEP (p<0.05) at T_lim. However, this was not associated with enhanced maximal voluntary activation during fatigue or recovery demonstrating that voluntary activation is not limited by central excitability.