Technical Abstract: Clinical assessment of infant growth and nutritional status is enhanced by accurate measurement of body composition and its changes over time. OBJECTIVE: To evaluate the accuracy of an air-displacement plethysmograph, the PEA POD® Infant Body Composition System (Life Measurement, Inc., Concord, CA), for determining percent body fat (%BF) and its changes over time against a criterion four-compartment (4C) model in an infant population. %BF by PEA POD, total body water by deuterium dilution, bone mineral content by dual-energy x-ray absorptiometry (DXA), and total body potassium by whole body counter were measured in 35 infants (2.7-23.0 wks) at their initial visit (baseline). 13 of these infants (2.7-12.6 wks) also completed a second visit after 1.5-mo. We compared %BF estimates from PEA POD and 4C model in 35 infants at baseline, and then examined whether PEA POD and 4C model detected similar %BF changes in the 13 infants who completed the follow-up visit. At baseline, mean %BF by PEA POD did not differ significantly from that by 4C model (17.7±5.4 vs. 18.1±6.7, P=0.60), while there was a significant difference between %BF by DXA and 4C model (22.5±7.7 vs. 18.1±6.7, P<0.01). The regression for %BF by 4C model vs. PEA POD did not significantly deviate from the line of identity (R2=0.7, SEE=4.0), while the regression for %BF by 4C model vs. DXA significantly deviated from the line of identity (R2=0.5, SEE=4.9). In the 13 infants who completed the follow-up visit, neither significant main effect of method (PEA POD vs. 4C model) (P=0.83) nor significant interaction between method and time (P=0.75) on %BF was detected, based on repeated measures ANOVA. This indicated that PEA POD and 4C model not only gave comparable estimates of %BF, but also measured longitudinal changes in %BF to the same extent and with similar sensitivity. However, %BF estimates from DXA and 4C model were significantly different from a cross-sectional comparison (P<0.01), although they detected similar changes in %BF (P=0.90). When compared to a 4C model, the PEA POD provides an accurate assessment of %BF and its changes over time in infants, and appears more accurate than DXA.