This study investigated (a) the feasibility and repeatability of intramuscular fine-wire electromyographic (fEMG) recordings from leg muscles during the repetitive, high-velocity cycling movement, (b) the influence of amplitude normalization technique on repeatability and statistical sensitivity, (c) the influence of test-retest interval duration on repeatability, and (d) differences between fEMG and surface EMG (sEMG) recordings of cycling. EMG activity of leg muscles was recorded using surface and fine-wire electrodes during one (n=12, to investigate statistical sensitivity and compare sEMG and fEMG) or two sessions (T1 and T2, 5-20 days apart, n=10, to investigate repeatability). fEMG recordings were feasible and there was high repeatability of fEMG recordings normalised to maximum measured EMG amplitude (MAX); mean coefficients of multiple correlation (CMC) ranged from .83+/-.13 to .88+/-.07. Data normalised to maximal (MVC) or submaximal contractions (sMVC) were less repeatable (p<.01). Statistical sensitivity was also greatest for data normalised to MAX (p<.01). Repeatability of fEMG increased with greater test-retest intervals (p<.01). The global pattern of muscle recruitment was consistent between sEMG and fEMG but sEMG recordings were characterized by additional myoelectric content. These findings support and guide the use of fEMG techniques to investigate leg muscle recruitment during cycling.