This study examined the reliability and accuracy of various velocity-based methods for predicting the snatch one-repetition maximum (1RM), with the purpose of evaluating the potential of movement velocity as an objective indicator for attempt selection in competitive weightlifting. Fourteen competitive adolescent male weightlifters (age: 15.9 ± 0.9 years; training experience: 5.1 ± 0.9 years) completed two testing sessions, each involving an incremental loading test with attempts at 50%, 70%, 80%, and 90% of their best snatch record from the past 30 days, followed by load increases until reaching their actual 1RM. Peak velocity (PV) was recorded for all lifts with a linear position transducer. The 1RM in the second session was predicted using the load-PV relationship derived from four loads, combined with either the actual or optimal minimal velocity threshold (MVT) obtained in the first session. Additionally, 1RM was estimated from PV recorded at single loads (50%, 70%, 80%, and 90% 1RM), using the individual %1RM-PV relationship established during the first session. Acceptable between-sessions reliability was observed for the actual 1RM, PV tested at single loads (50-90% 1RM), actual MVT, and optimal MVT (intraclass correlation coefficient = 0.76-0.90, coefficient of variation = 1.82-3.31%). The actual MVT, optimal MVT and individual %1RM-PV relationship using 80% and 90% 1RM yielded acceptable and lower absolute errors (2.6-4.1 kg) compared to individual %1RM-PV relationship using 50% and 70% 1RM (6.2-9.9 kg). However, these methods exhibited proportional bias (p = 0.002-0.018). Furthermore, heteroscedasticity was observed for the actual MVT and 90% 1RM methods (p = 0.022-0.026). These results suggest that recording PV during warm-up sets prior to competition may serve as a complementary variable to support and refine opener selection in weightlifting competitions. However, weightlifting coaches should use this approach with caution, as it may not provide accurate snatch performance predictions for all athletes.

• The actual MVT, optimal MVT, and PVs at submaximal loads showed acceptable between-session reliability.

• Predicted 1RMs showed no systematic error, with absolute errors below 5 kg when using actual MVT, optimal MVT, and the %1RM-PV relationship with 80% and 90% 1RM.

• Above methods showed proportional bias, and for actual MVT and 90% 1RM methods, prediction errors increased with athlete performance (heteroscedasticity).