The well-known ergogenic effect of caffeine on submaximal exercise performance is caused largely by a reduction in perception of effort (Doherty & Smith, 2005). However, the brain mechanisms underlying this effect are poorly understood. Perception of effort is thought to arise from corollary discharges of the central motor command to the active muscles (McCloskey et al., 1983), and there is evidence that movement-related cortical potential (MRCP) derived from the EEG is associated with central motor command (Siemionow et al., 2000; Jankelowitz & Colebatch, 2005), and perception of effort (Slobounov et al., 2004). Our aim was to investigate whether caffeine affects MRCP amplitude and rating of perceived effort (RPE) during submaximal isometric knee-extensor contractions, and whether caffeine affects brain activity related to movement planning and/or brain activity related to movement execution. Twelve healthy, recreationally active women (age, mean 21 ± standard deviation 5 years, height 166 ± 7 cm, weight 67 ± 11 kg) performed 100 isometric knee-extensor contractions at 61 ± 5% of their maximal voluntary contraction torque 1.5 hours after caffeine (6 mg/kg) or alternative placebo ingestion, while RPE, vastus lateralis (VL) EMG, and MRCP were recorded. Conditions were presented in a randomly counterbalanced and double blind fashion. The effect of exercise duration was assessed by comparing the first 50 with the last 50 contractions. Two-way fully repeated measures ANOVAs (caffeine x exercise duration) showed that caffeine caused a significant reduction in RPE (caffeine 5.3 ± 1.7, placebo 5.9 ± 1.7; F(1,11) = 8.51, p = 0.014, ηp2 = 0.44), and that exercise duration caused a significant increase in RPE (first block 4.8 ± 1.7, second block 6.4 ± 1.6; F(1,11) = 16.32, p = 0.002, ηp2 = 0.60). Follow-up tests of the four-way fully repeated measures ANOVA (electrode x epoch x caffeine x exercise duration) for MRCP showed that MRCP amplitude at the vertex during the first second of movement execution was significantly decreased by caffeine (caffeine −12.1 ± 6.5 µV, placebo −16.2 ± 5.4 µV; t(11) = 4.44, p = 0.001, ηp2 = 0.64). A planned comparison showed a significant increase in the same MRCP component with exercise duration (first block −13.4 ± 6.0 µV, second block −14.9 ± 5.6 µV; t(11) = 2.97, p = 0.013, ηp2 = 0.45). Contraction torque and VL EMG were not affected by caffeine and exercise duration. These results demonstrate for the first time that the positive effect of caffeine on perception of effort is associated with a reduction in the central motor command required to produce the same motor neuron output and force output. Opposite effects were observed for exercise duration. Furthermore, we have shown that perception of effort is associated with brain activity related to movement execution, not movement planning.