Abstract Purpose Most studies examining the predictive value of the load–velocity relationship in determining one-repetition maximum (1RM) in the back squat implemented its direct determination to enable testing movement velocity within a predetermined set of relative loads (e.g., 50, 60, 70, 80, and 90% 1RM). We determined whether a different approach of load–velocity profiling affects the accuracy of estimating 1RM. Methods Predictions based on a practical 2-point approach (no pre-determination of 1RM) were compared to those obtained with the conventional multipoint and 2-point approach (pre-determination of 1RM). 1RM was estimated relying on a reference minimum velocity threshold (MVT) of 0.3 ms−1. Analyses were conducted for separate back-squat variations (n = 13 Smith machine; n = 13 free-weight). Slopes and y-intercepts were compared. The accuracy of 1RM prediction was determined by contrasting actual vs. predicted 1RM values. Results The individual MVT did not differ from the general 0.3 ms−1 value in either back-squat variation. Slopes and y-intercepts were similar between all determination approaches. For the Smith machine, estimated 1RM did not differ from the actual value with either approach (mean misestimate: −1.83 to 0.02 kg). However, the limits of agreement were wide (~ 12 kg) and the absolute percent error was significantly different from 0 with all approaches (p < 0.05). Conclusion 1RM can be estimated with similar accuracy with all profiling methods, irrespectively of the back squat variation. However, the free-weight variation displays higher systematic and random errors. It can be concluded that the wide limits of agreement preclude accurate 1RM estimations on an individual basis.