the measurement precision of body composition can interfere on the diagnosis and prescription of diseases' treatment. Furthermore, with regard to dual energy X-rays absorptiometry (DXA), there may be important differences between the measures of regions of interest (ROI) automatically performed by DXA or manually by an evaluator, which can cause measurement error and influence the evaluation or diagnosis. thus, this study aimed to evaluate the measurement reliability of body composition by DXA and intraobserver reproducibility for the ROI measurement. a total of 15 young adults were subjected to two full-body scans by DXA, under the same conditions. The first scan of all volunteers was chosen and the extent of the ROI was performed, in two stages, separated by a period of 2 weeks, by a single trained evaluator. The coefficient of variation (CV) and intraclass correlation coefficient (ICC) were calculated with a significance level of p < 0.05. high-precision measurements of DXA for whole body mass (ICC = 0.999; CV = 2.3%), fat mass (ICC = 0.998; CV = 1.6%), lean mass (ICC = 0.995; CV = 0.3%) and bone mineral content (ICC = 0.996; CV < 0.1%) were obtained. Further, it was observed high intraobserver reproducibility for ROI measurement, with ICC values ranging between 0.952 and 0.999. body composition measurement by DXA presents high reproducibility for whole body mass, fat mass, lean mass and bone mineral content and also high intraobserver reproducibility for the ROI measurement.