Quantitative nuclear magnetic resonance (QMR) is being used in human adults to obtain measures of total body fat mass (FM) with high precision. The current study assessed a device specially designed to accommodate infants and children between 3 and 50 kg (EchoMRI-AH). Body composition of 113 infants and children (3.3-49.9 kg) was assessed using dual-energy X-ray absorptiometry (DXA), air displacement plethysmography (ADP, PeaPod for infants ≤ 8 kg and BodPod for children ≥ 6 years) and QMR. Results were compared with the deuterium oxide dilution technique (D(2)O) and a four-compartment model (4-C). The percentages of compliance were: 98% QMR; 75% DXA; 94% BodPod; and 95% PeaPod. Although QMR precision was high (coefficient of variation = 1.42%), it overestimated FM ~10% compared to the 4-C model and underestimated FM by ~4% compared to the deuterium method in children ≥ 6 years. QMR was less concordant with 4-C or D(2)O models for infants ≤ 8 kg. Thus, a piece-wise defined model for mathematically fitting the QMR data to the D(2)O data was employed and this adjustment improved the accuracy relative to D(2)O and 4-C for infants. Our results suggest that the pediatric QMR with appropriate mathematical adjustment provides a fast and precise method for assessing FM longitudinally in infants and in children weighing up to 50 kg.