Peak oxygen uptake (V̇O2peak) is routinely expressed in litres per minute and by unit of body mass (mL·kg(-1)·min(-1)) despite the theoretical and statistical limitations of using ratios. Allometric modeling is an effective approach for partitioning body-size effects in a performance variable. The current study examined the relationships among chronological age (CA), skeletal age (SA), total body and appendicular size descriptors, and V̇O2peak in male adolescent roller hockey players. Seventy-three Portuguese, highly trained male athletes (CA, 15.4 ± 0.6 years; SA, 16.4 ± 1.5 years; stature, 169.9 ± 6.9 cm; body mass, 63.7 ± 10.7 kg; thigh volume, 4.8 ± 1.0 L) performed an incremental maximal test on a motorized treadmill. Exponents for body size descriptors were 2.15 for stature (R(2) = 0.30, p < 0.01) and 0.55 for thigh volume (R(2) = 0.46, p < 0.01). The combination of stature or thigh volume and CA or SA, and CA(2) or SA(2), increased the explained variance in V̇O2peak (R(2) ranged from 0.30 to 0.55). The findings of the allometric model combining more than 1 body size descriptor (i.e., stature and thigh volume) in addition to SA and CA(2) were not significant. Results suggest that thigh volume and SA are the main contributors to interindividual variability in aerobic fitness.