The primary aim of this study was to explore the goodness of fit and accuracy of both general and individual relationships between the magnitude of velocity loss (%VL) and the percentage of performed repetitions with respect to the maximal number of repetitions that can be completed to failure (%Rep) during the Smith machine prone bench pull exercise. Fifteen male sports science students completed a preliminary session to determine the bench pull one-repetition maximum (1RM) and 2 identical experimental sessions separated by 48–72 hours. In each experimental session, subjects randomly performed single sets of repetitions to failure separated by 10 minutes against the 60% 1RM, 70% 1RM, and 80% 1RM during the Smith machine bench pull exercise. Individual %Rep-%VL relationships presented a greater goodness of fit than general %Rep-%VL relationships at the 60% 1RM (R2 = 0.85–0.97 vs. 0.79–0.85), 70% 1RM (R2 = 0.84–0.99 vs. 0.77–0.84), and 80% 1RM (R2 = 0.84–1.00 vs. 0.74–0.80). However, the accuracy (absolute errors) in estimating the %Rep during the second testing session based on the %Rep-%VL equations obtained in the first testing session did not differ between the individual and general %Rep-%VL equations in 8 of 9 comparisons (p ≥ 0.102). The absolute errors between the actual and predicted %REP were unacceptable (>10%) in 11 of 18 comparisons, and acceptable (5–10%) in 7 of 18 comparisons. These results highlight that the %Rep cannot be estimated with high degree of accuracy from VL recordings during the Smith machine bench pull exercise, regardless of whether individual or general Rep-%VL relationships are considered.