The objective of the study was to determine the precision of total- and regional-body composition measurements from a total-body scan using dual-energy X-ray absorptiometry (DXA). This is critical information necessary to determine the smallest change from baseline that could be detected with statistical significance when conducting longitudinal measurements of body composition variables in an individual. Twenty volunteers were scanned once each day for 4 consecutive days using a Lunar DPX-L densitometer and manufacturer-supplied software (version 1.3z). Coefficients of variation (CV, %) derived from data using the (preferred) extended research mode of analysis were 0.62, 1.89, 0.63, 2.0, 1.11, 1.10, and 1.09% for total-body bone mineral density (BMD), total percentage fat, total body tissue mass, fat mass, lean mass, bone mineral content (BMC), and total bone calcium, respectively. Regional measurements (arm, leg, trunk, pelvis, and spine) were less precise than total body measurements, with CVs in the range of 1% to 3% (but fat mass for arms was 4.26%, trunk 3.08%, BMC 3.65%). Small but statistically significant differences in mean values for most body composition variables were found when data were compared between extended and standard modes of analysis. Inconsistent use of analysis mode in a cohort or when following a patient longitudinally may negatively affect precision. We conclude that the measurement precision of total and regional body composition variables was generally comparable to the precision limits typically associated with lumbar spine and proximal femur BMD data.