This study examined the accuracy of different velocity-based methods in the prediction of bench press and squat one-repetition maximum (1RM) in female athletes. Seventeen trained females (age 17.8 ± 1.3 years) performed an incremental loading test to 1RM on bench press and squat with the mean velocity being recorded. The 1RM was estimated from the load–velocity relationship using the multiple- (8 loads) and two-point (2 loads) methods and group and individual minimum velocity thresholds (MVT). No significant effect of method, MVT or interaction was observed for the two exercises (p > 0.05). For bench press and squat, all prediction methods demonstrated very large to nearly perfect correlations with respect to the actual 1RM (r range = 0.76 to 0.97). The absolute error (range = 2.1 to 3.8 kg) for bench press demonstrated low errors that were independent of the method and MVT used. For squat, the favorable group MVT errors for the multiple- and two-point methods (absolute error = 7.8 and 9.7 kg, respectively) were greater than the individual MVT errors (absolute error = 4.9 and 6.3 kg, respectively). The 1RM can be accurately predicted from the load–velocity relationship in trained females, with the two-point method offering a quick and less fatiguing alternative to the multiple-point method.