Purpose: To compare the load–velocity (L-V) relationship between bench-press exercises performed using 4 different grip widths, to determine the association between the anthropometric characteristics and L-V profile, and to explore whether a multiple linear-regression model with movement velocity and subjects’ anthropometric characteristics as predictor variables could increase the goodness of fit of the individualized L-V relationship. Methods: The individual L-V relationship of 20 men was evaluated by means of an incremental loading test during the bench-press exercise performed on a Smith machine using narrow, medium, wide, and self-selected grip widths. Simple and multiple linear-regression models were performed. Results: The mean velocity associated with each relative load did not differ among the 4 grip widths (P ≥ .130). Only body height and total arm length were correlated with the mean velocity associated with light and medium loads (r ≥ .464). A slightly higher variance of the velocity attained at each relative load was explained when some anthropometric characteristics were used as predictor variables along with the movement velocity (r² = .969 [.965–.973]) in comparison with the movement velocity alone (r² = .966 [.955–.968]). However, the amount of variance explained by the individual L-V relationships was always higher than with the multiple linear-regression models (r² = .995 [.985–1.000]). Conclusions: These results indicate that the individual determination of the L-V relationship using a self-selected grip width could be recommended to monitor relative loads in the Smith machine bench-press exercise.