The purpose of this study was to determine the reliability of three normalization methods for analyzing hip abductor activation during rehabilitation exercises. Thirteen healthy subjects performed three open kinetic chain and three closed kinetic chain hip abductor exercises. Surface EMG activity for the gluteus medius was collected during each exercise and normalized based on a maximum voluntary isometric contraction (MVIC), mean dynamic (m-DYN), and peak dynamic activity (pk-DYN). Intraclass coefficient correlations (ICCs), intersubject coefficients of variation (CVs), and intrasubject CVs were then calculated for each normalization method. MVIC ICCs exceeded 0.93 for all exercises. M-DYN and pk-DYN ICCs exceeded 0.85 for all exercises except for the sidelying abduction exercise. Intersubject CVs ranged from 55% to 77% and 19% to 61% for the MVIC and dynamic methods, respectively. Intrasubject CVs ranged from 11% to 22% for all exercises under all normalization methods. The MVIC method provided the highest measurement reliability for determining differences in activation amplitudes between hip abductor exercises in healthy subjects. Future research should determine if these same results would apply to a symptomatic patient population.