Reproducibility of frequency content from surface electromyography (sEMG) signals has not been assessed and it is unknown if incremental load testing design could affect sEMG in cycling. The goals of this study were to assess the reproducibility of measures from sEMG frequency content between sessions and to compare these frequency components between a ramp and a step incremental cycling test. Eighteen cyclists performed four incremental load cycling tests to exhaustion. Two tests were performed using a step increment (load started at 100 W for 3 min followed by increments of 30 W every 3 min) and two were performed using a ramp increment (load started at 100 W for 1 min followed by increments of 30 W·min-1). sEMG was monitored bilaterally for the rectus femoris and vastus lateralis throughout the tests and converted into overall activation (whole signal bandwidth), high- and low-frequency contents. The reproducibility of the frequency content ranged from none to strong (ICC = 0.07-0.90). Vastus lateralis activation was larger at the step compared to the ramp test (P < 0.01), without differences for rectus femoris (P = 0.22-0.91) and for the high-frequency (P = 0.28-0.95) and low-frequency contents (P = 0.13-0.94). sEMG from vastus lateralis and rectus femoris presented none to strong reproducibility. Vastus lateralis is more activated in step test design.