Arm cycling is commonly used in rehabilitation settings for individuals with motor impairments in an attempt to facilitate neural plasticity, potentially leading to enhanced motor function in the affected limb(s). Studies examining the neural control of arm cycling, however, typically cycle using a set cadence and power output. Given the importance of motor output intensity, typically represented by the amplitude of electromyographic (EMG) activity, on neural excitability, surprisingly little is known about how arm muscle activity is modulated using relative workloads. Thus, the objective of this study was to characterize arm muscle activity during arm cycling at different relative workloads. Participants ( = 11) first completed a 10-second maximal arm ergometry sprint to determine peak power output (PPO) followed by 11 randomized trials of 20-second arm cycling bouts ranging from 5-50% of PPO (5% increments) and a standard 25 W workload. All submaximal trials were completed at 60 rpm. Integrated EMG amplitude (iEMG) was assessed from the biceps brachii, brachioradialis, triceps brachii, flexor carpi radialis, extensor carpi radialis and anterior deltoid of the dominant arm. Arm cycling was separated into two phases, flexion and extension, relative to the elbow joint for all comparisons. As expected, iEMG amplitude increased during both phases of cycling for all muscles examined. With the exception of the triceps brachii and extensor carpi radialis, iEMG amplitudes differed between the flexion and extension phases. Finally, there was a linear relationship between iEMG amplitude and the %PPO for all muscles during both elbow flexion and extension.