This study examined post-exercise changes in corticospinal excitability in five 'elite' rowers and six non-rowers. Transcranial magnetic stimulation (TMS) was delivered to the motor cortex and bilateral electromyographic (EMG) recordings were made from erector spinae (ES) muscles at L3/L4 spinal level and from the first dorsal interosseous (FDI) muscle of the dominant hand. Each subject completed two exercise protocols on a rowing ergometer: a light exercise protocol at a sub-maximal output for 10 min and an intense exercise protocol at maximum output for 1 min. A trial of ten magnetic stimuli was delivered before each of the protocols and, on finishing exercise, further trials of ten stimuli were delivered every 2 min for a 16 min period. Amplitudes of motor-evoked potentials (MEPs) in each of the three test muscles were measured before exercise and during the recovery period after exercise. The non-rowers showed a brief facilitation of MEPs in ES 2 min after light and intense exercise that was only present in the elite rowers after intense exercise. In the period 4-16 min after light exercise, the mean (± S.E.M.) MEP amplitude (relative to pre-exercise levels) was less depressed in the elite rowers (79.4 ± 2.1 %) than in the non-rowers (60.9 ± 2.5 %) in the left ES but not significantly so in the right ES. MEP amplitudes in FDI were significantly larger in the elite rowers, averaging 119.0 ± 3.1 % pre-exercise levels, compared with 101.2 ± 5.8 % in the non-rowers. Pre-exercise MEP latencies were no different in the two groups. After light exercise MEP latencies became longer in the elite rowers (left ES, 16.1 ± 0.5 ms; right ES, 16.1 ± 0.4 ms; dominant FDI, 23.4 ± 0.2 ms) than in the non-rowers (left ES, 15.0 ± 0.3 ms; right ES, 15.2 ± 0.3 ms; dominant FDI, 21.5 ± 0.2 ms). There were no differences in MEP depression or latency between elite rowers and non-rowers after intense exercise. We conclude that the smaller degree of MEP depression in the elite rowers after light exercise reflects less central fatigue within corticospinal control pathways than that seen in the non-rowers. The longer latency of MEPs seen in the elite rowers may reflect recruitment of more slower-conducting fatigue-resistant motor units compared with the non-rowers. These differences may be because the energy requirements for the non-rowers during light exercise are closer to their maximum capacity, leading to more fatigue. This notion is supported by the lack of any difference between groups following intense exercise when both groups were working at their own maximum. Experimental Physiology (2002) 87.5, 593-600.