We hypothesized that if reduced spinal excitability contributes to central activation failure, then a caffeine-induced increase in spinal excitability would enhance postfatigue maximal voluntary activation and maximal voluntary contraction (MVC). Ten male volunteer subjects attended two laboratory sessions separated by at least 1 week. Contractile and electrical properties were assessed before, and 1 h after oral administration of caffeine (6 mg/kg) or placebo (all-purpose flour), and again following a fatigue protocol. The slope of the H reflex recruitment curve, normalized to that of the M wave (H(slp)/M(slp)), was used to estimate spinal excitability. Maximal voluntary activation was assessed using maximal EMG (EMG(max)) and twitch interpolation. Postfatigue, MVC torque declined (P<0.05) to 75.2+/-12.7 and 70.2+/-9.3% of the prefatigue values in the placebo (PL) and caffeine (CF) trials, respectively, and remained depressed throughout the recovery period. This was accompanied by a decline in % activation (P<0.05) from 99.6+/-0.3% (PL) and 99.8+/-0.3% (CF) to 94.8+/-3.5% (PL) and 95.3+/-5.0% (CF), indicating the presence of central activation failure. Caffeine offset the decline in H(slp)/M(slp )observed in the placebo trial (P<0.05), but it did not prevent the decline in maximal voluntary activation or MVC torque. Furthermore, although the decline in spinal excitability was correlated to the decline in EMG(max) (r=0.55, P<0.05) it was not correlated with the decline in % activation or MVC torque. Thus a fatigue-induced decline in spinal excitability did not limit maximal activation.