Introduction Resistance and reactance are often used in body composition compartment regressions which require the assumptions that the body is a cylinder of constant transversal area and the hydration is constant. Problems can be found in extremes of body composition, when those assumptions are not met.

Objective To analyse if the impedance equation estimative of body fat (BF) agrees with that provided by the Dexa reference method.

Methods We used representative data of the North American population, entitled Nhanes 2003–3004. Individuals aged 20–49 from both sexes (n=1716) were selected and information on BF% estimated by Dexa, resistance, reactance, height and weight were used. Impedance equation was proposed by Kyle et al for lean body mass: −4.104+(0.518* Height²/Resistance)+(0.231* weight)+(0.130* Reactance)+(4.229×sex); Sex: man=1 and woman=0. Weight minus lean mass provided the BF%. BF% was divided in four categories: 15%, 15%–25%, 25%–35%, and 35%–45%. κ Statistic was used for evaluating agreement between both methods, in each category of BF%, in each sex.

Results κ Statistics from lowest to highest categories of BF% were 0.35; 0.38; 0.47; 0.46 and 0.39; 0.51; 0.48 and 0.63 for male and female, respectively (all p<0.001).

Conclusion Estimates of %BF by impedance equation and Dexa differ, mainly in lowest categories of %BF and among males, but not for highest category among females. Caution must be taken in using such equations among individuals with extreme body composition compartments.