This study examined the accuracy of predicting a free-weight back squat and a bench press one-repetition maximum (1RM) using both 2- and 4-point submaximal average concentric velocity (ACV) methods. Seventeen resistance trained men performed a warm-up and a 1RM test on the squat and bench press with ACV assessed on all repetitions. The ACVs during the warm-up closest to 1.0 and 0.5m.s-1 were used in the 2-point linear regression forecast of the 1RM and the ACVs established at loads closest to 20, 50, 70, and 80% of the 1RM were used in the 4-point 1RM prediction. Repeated measures ANOVA and Bland-Altman and Mountain plots were used to analyze agreement between predicted and actual 1RMs. ANOVA indicated significant differences between the predicted and the actual 1RM for both the 2- and 4-point equations in both exercises (p<0.001). The 2-point squat prediction overestimated the 1RM by 29.12±0.07kg and the 4-point squat prediction overestimated the 1RM by 38.53±5.01kg. The bench press 1RM was overestimated by 9.32±4.68kg with the 2-point method and by 7.15±6.66kg using the 4-point method. Bland-Altman and Mountain plots confirmed the ANOVA findings as data were not tightly conformed to the respective zero difference lines and Bland-Altman plots showed wide limits of agreement. These data demonstrate that both 2- and 4-point velocity methods predicted the bench press 1RM more accurately than the squat 1RM. However, a lack of agreement between the predicted and the actual 1RM was observed for both exercises when volitional velocity was used.