Training at the load that maximizes peak mechanical power (Pmax) is considered superior for the development of power. We aimed to identify the Pmax load ('optimal load') in the jump squat and to quantify small, moderate, large, and very large substantial differences in power output across a spectrum of loads to identify loads that are substantially different to the optimal, and lastly, to investigate the nature of power production (load-force-velocity profiles). Professional Australian Rules Football (ARF; n = 16) and highly trained Rugby Union (RU; n = 20) players (subdivided into stronger [SP] vs. weaker [WP] players) performed jump squats across incremental loads (0-60 kg). Substantial differences in peak power (W·kg(-1)) were quantified as 0.2-2.0 of the log transformed between-athlete SD at each load, backtransformed and expressed as a percent with 90% confidence limits (CL). A 0-kg jump squat maximized peak power (ARF: 57.7 ± 10.8 W·kg(-1); RU: 61.4 ± 8.5 W·kg(-1); SP: 64.4 ± 7.5 W·kg(-1); WP: 54.8 ± 9.5 W·kg(-1)). The range for small to very large substantial differences in power output was 4.5-55.9% (CL: ×/÷1.36) and 2.8-32.4% (CL: ×/÷1.31) in ARF and RU players, whereas in SP and WP, it was 3.7-43.1% (CL: ×/÷1.32) and 4.3-51.7% (CL: ×/÷1.36). Power declined per 10-kg increment in load, 14.1% (CL: ±1.6) and 10.5% (CL: ±1.5) in ARF and RU players and 12.8% (CL: ±1.9) and 11.3% (CL: ±1.7) in SP and WP. The use of a 0-kg load is superior for the development of jump squat maximal power, with moderate to very large declines in power output observed at 10- to 60-kg loads. Yet, performance of heavier load jump squats that are substantially different to the optimal load are important in the development of sport-specific force-velocity qualities and should not be excluded.