Scaling left ventricular (LV) mass and other cardiac dimensions to account for individual body size is important. The traditional method of simple ratio scaling using, for example, body surface area (BSA) assumes a linear and proportional relationship and accurate measurement of both LV mass and BSA. These assumptions can be questioned; hence, we examined the appropriateness of methods and different indices using highly accurate magnetic resonance imaging scans. Cardiac and whole-body scans were performed in 172 young, healthy, male subjects (age range, 17-28 years) to assess LV mass, volume, linear dimensions, lean body mass and fat mass. Height, body mass and BSA were determined anthropometrically. Relationships were examined for linearity and closeness of fit using log-log least-squares linear regression to determine the slope exponent b (where 1.0 indicates linearity). The relationship between LV mass and lean body mass (b=.90+/-.15; r(2)=.66) was linear and geometrically consistent. This was also the case for LV end-diastolic volume (b=.70), although the confidence intervals were broader (+/-0.32) and the r(2) (.31) smaller. The relationships between LV mass, volume and other variables were generally not linear or geometrically consistent. LV linear dimensions did not demonstrate any linear relationships, and in particular, those with BSA were extremely poor (r(2)=.02-.09). In summary, the traditional scaling of LV measurements to BSA does not remove the influence of body size and other techniques should be considered. Lean body mass was the most appropriate variable for simple indexing of LV mass. No body size variable had a linear and proportional relationship with LV linear dimensions, and the use of simple ratio scaling for these is seriously questioned.