Fat-free mass (FFM) has been proposed as an optimal normalization of left ventricular (LV) mass to body size. We sought to evaluate the novel FFM-based criteria of LV hypertrophy (LVH). A population sample of 1371 men and women aged 25 to 74 years was examined by echocardiography and bioelectrical impedance analysis. Internal partition values for LVH were generated in a healthy population subgroup on the basis of LV mass divided by FFM and by the traditional indexations to body height, height(2.7), and body surface area. In contrast to the sex-specific criteria required by traditional indexations, the value of LV mass/FFM that divided individuals with and without LVH was identical for men and women (4.1 g/kg). Estimates of LVH prevalence varied significantly by type of indexation used, internally or externally derived cut points, and by population subgroups. Differences were pronounced among hypertensives and the obese. Thus, the application of LV mass/FFM more than halved the risk of LVH in obese versus nonobese women (odds ratio, 2.5; 95% confidence interval, 1.6 to 4.0) compared with criteria based on LV mass/height(2.7) (odds ratio, 5.5; 95% confidence interval, 3.6 to 8.3). Implications among hypertensives were less marked. Indexation of LV mass to FFM eliminates sex-specific LVH criteria. The proportion of individuals defined as having LVH using the new criteria deviate markedly from traditional indexations. Prospective investigations will be needed to identify the prognostic implications of different indexations, especially in subgroups such as the obese.