This study aimed to evaluate whether lifting velocity can be used to estimate the overhead press one repetition maximum (1RM) and to explore the differences in the accuracy of the 1RM between three velocity-based methods. Twenty-seven weightlifters (16 men and 11 women) participated. The first session was used to test the overhead press 1RM. The second session consisted of an incremental loading test during the overhead press. The mean velocity was registered using a transducer attached to the barbell. A 1-way repeated-measures analysis of variance (ANOVA) with Bonferroni post hoc corrections was applied to the absolute differences between the actual and predicted 1RMs. Raw differences with 95% limits of agreement and ordinary least-products regressions were used to test the concurrent validity of the 1RM prediction methods with respect to the actual 1RM. The ANOVA did not reveal significant differences for the absolute differences respect to the actual 1RM between the three 1RM prediction methods ( F = 3.2, p = .073). The absolute errors were moderate for the Multiple-Point (6.1 ± 3.7%), Two-Point₄₅₋₇₅ (8.6 ± 6.2%), and Two-Point₄₅₋₉₀ methods (5.7 ± 4.0%). The validity analysis showed that all the 1RM prediction methods underestimated the actual 1RM (1.0–2.2 kg), but ordinary least-products regressions failed to show fixed or proportional bias. These results suggest that the Multiple-Point and Two-Point₄₅₋₉₀ velocity-based methods might be viable tools to predict the overhead press 1RM in weightlifters, but practitioners are encouraged to use the direct 1RM for a more accurate prescription of the training loads.