This study investigated different methods of scaling submaximal cardiac output (Q) and stroke volume (SV) to best normalize for body size (body surface area [BSA], height [Ht], weight [Wt], and fat-free mass [FFM]). Q and SV were measured at both an absolute (50 W) and a relative power output (60 % of VO2max) in 337 men and 422 women, 17 to 65 years of age. Traditional ratio scaling was examined in addition to allometric scaling, where scaling exponents ( B) were determined for each body size variable (x) that best normalized the physiological outcome variables (y) for body size (y = ax(b)). With ratio scaling, regardless of the body size variable (x = BSA, Ht, Wt, FFM), there was no evidence of a linear relationship between x and y (y = Q or SV). A linear relationship is a necessary condition for appropriate normalization. Further, when ratio-scaled variables (e.g., Q/BSA) were correlated to the body size variable (e.g., BSA) by which they were scaled, significant (p <or= 0.05) relationships still existed for BSA, Ht, Wt, and FFM. Thus, ratio scaling did not meet either criteria for normalizing Q and SV for body size. In contrast, when allometrically-derived scaling exponents were used to normalize Q and SV (e.g., Q/BSA(b)), the resulting scaled values were uncorrelated (i.e., size-independent) with BSA, Ht, Wt, or FFM. These results were independent of age, sex or race. In summary, ratio scaling did not appropriately normalize Q and SV for differences in body size, while allometric scaling did result in size-independent values. Thus, individually-derived allometric exponents should be applied to body size variables to most appropriately adjust Q and SV for body size.