Background This study aimed to identify which maximum voluntary isometric contraction (MVIC) and sub-MVIC tests produce the highest activation of the erector spinae muscles and the greatest reduction in inter-individual variability, to put them forward as reference normalization maneuvers for future studies. Methods Erector spinae EMG activity was recorded in 38 healthy women during five submaximal and three maximal exercises. Results None of the three MVIC tests generated the maximal activation level in all the participants. The maximal activation level was achieved in 68.4% of cases with the test performed on the roman chair in the horizontal position (96.3 ± 7.3; p < 0.01). Of the five submaximal maneuvers, the one in the horizontal position on the roman chair produced the highest percentage of activation (61.1 ± 16.7; p < 0.01), and one of the lowest inter-individual variability values in the normalized signal of a trunk flexion-extension task. Conclusions A modified Sorensen MVIC test in a horizontal position on a roman chair and against resistance produced the highest erector spinae activation, but not in 100% of participants, so the execution of several normalization maneuvers with the trunk at different inclinations should be considered to normalize the erector spinae EMG signal. A modified Sorensen test in a horizontal position without resistance is the submaximal maneuver that produces the highest muscle activation and the greatest reduction in inter-individual variability, and could be considered a good reference test for normalization.