Neuromuscular activation is considered an important determinant sprint cycling performance but requires reliable EMG amplitude measurements to facilitate sensitive assessments. The reliability of EMG measurements during sprint cycling may depend on the sprint cycling test undertaken (isovelocity or isoinertial accelerating), the reference tasks used for normalisation (isometric MVCs of a series of single muscle groups [ISO-SINGJT] or isometric cycling MVCs [ISO-CYC]), and the efficacy of the normalisation. This study aimed to compare the magnitude and between-session reliability of peak muscle activation (peak rmsEMG) during: isovelocity and isoinerital sprint cycling tests; ISO-SINGJT and ISO-CYC reference tasks; and absolute and normalised EMG during the sprint cycling tests. EMG amplitude was measured over six major muscle groups on both legs and all measurements were made over two sessions in a randomised counterbalanced design. Peak rmsEMG was assessed during both ISO-SINGJT and ISO-CYC MVCs and then during mechanical peak power output (PPO) during isovelocity (120 RPM) and isoinerital acceleration (0 to >150RPM) sprint tests. Absolute peak rmsEMG and for the sprint tests normalised EMG values were determined, and coefficient of variation and intra-class correlation coefficients used to assess reliability. Peak rmsEMG at PPO during both sprint cycling tests was similar for the six muscle groups measured. Peak rmsEMG was higher during ISO-SINGJT than ISO-CYC for for 3 of the 6 muscle groups, but all muscle groups exhibited similar reliability for both reference tasks. Neither reference task improved the between-session reliability for either sprint test. This data highlights reservations in the use of isometric reference tasks to ascertain changes in peak muscle activation over time in during sprint cycling assessments.