AbstractThis study aimed to analyze the effects of three different velocity loss thresholds (10%: VL10, 30%: VL30 and 50%: VL50) on maximal strength and velocity at different intensities during bench press using a weight stack machine (WSM-BP). Forty-five men were randomly assigned to three groups: VL10 (n=15), VL30 (n=15) and VL50 (n=15), which followed a 5-week (15 sessions) velocity-based WSM-BP program. Assessments performed Pre- and Post-training included: a) estimated one-repetition maximum (1RM) in WSM-BP; b) average velocity attained against all absolute loads common to Pre- and Post-training tests in WSM-BP; c) average velocity attained against all absolute loads that were lifted equal to or faster than 0.8 m·s−1 at Pre-training (light loads); and d) average velocity attained against all absolute loads that were lifted slower than 0.8 m·s−1 at Pre-training (heavy loads). All groups showed significant improvements in 1RM, velocity against all loads, and velocity against heavy loads (P<0.001–0.01). However, only the VL10 group showed significant enhancements in velocity against light loads (P=0.05). Therefore, the VL10 group showed a higher training efficiency compared to VL30 and VL50 interventions, since it obtained similar benefits by performing fewer repetitions.