This study aimed to compare the between-session reliability of different performance variables during 2 variants of the Smith machine back-squat exercise. Twenty-six male wrestlers performed 5 testing sessions (a 1-repetition maximum [1RM] session, and 4 experimental sessions [2 with the pause and 2 with the rebound technique]). Each experimental session consisted of performing 3 repetitions against 5 loads (45–55–65–75–85% of the 1RM). Mean velocity (MV), mean power (MP), peak velocity (PV), and peak power (PP) variables were recorded by a linear position transducer (GymAware PowerTool). The best and average scores of the 3 repetitions were considered for statistical analyses. The coefficient of variation (CV) ranged from 3.89% (best PV score at 55% 1 RM using the pause technique) to 10.29% (average PP score at 85% 1 RM using the rebound technique). PP showed a lower reliability than MV, MP, and PV (CVratio ≥ 1.26). The reliability was comparable between the exercise techniques (CVratio = 1.08) and between the best and average scores (CVratio = 1.04). These results discourage the use of PP to assess back-squat performance at submaximal loads. The remaining variables (MV, MP, or PV), exercise techniques (pause or rebound), and repetition criteria (best score or average score) can be indistinctly used due to their acceptable and comparable reliability.