This study compared peak power estimated using 4 commonly used regression equations with actual peak power derived from force platform data in a group of adolescent basketball players. Twenty-five elite junior male basketball players (age, 16.5 +/- 0.5 years; mass, 74.2 +/- 11.8 kg; height, 181.8 +/- 8.1 cm) volunteered to participate in the study. Actual peak power was determined using a countermovement vertical jump on a force platform. Estimated peak power was determined using countermovement jump height and body mass. All 4 prediction equations were significantly related to actual peak power (all p < 0.01). Repeated-measures analysis of variance indicated significant differences between actual peak power and estimate peak power from all 4 prediction equations (p < 0.001). Bonferroni post hoc tests indicated that estimated peak power was significantly lower than actual peak power for all 4 prediction equations. Ratio limits of agreement for actual peak power and estimated peak power were 8% for the Harman et al. and Sayers squat jump prediction equations, 12% for the Canavan and Vescovi equation, and 6% for the Sayers countermovement jump equation. In all cases peak power was underestimated.