MRI is finding increased clinical use in neonatal populations; the extent to which electromagnetic models used for quantification of specific absorption rate (SAR) by commercial MRI scanners accurately reflect this alternative scenario is unclear. This study investigates how SAR predictions relating to adults can be related to neonates under differing conditions when imaged using 1.5 T and 3 T MRI scanners. Electromagnetic simulations were produced in neonatal subjects of different sizes and positions within a generic MRI body transmit device operating at both 64 MHz and 128 MHz, corresponding to 1.5 T and 3 T MRI scanners, respectively. An adult model was also simulated, as was a spherical salt‐water phantom, which was also used in a calorimetry experiment. The SAR in neonatal subjects was found to be less than that experienced in an adult in all scenarios; however, the overestimation factor was variable. For example a 3 T body scan resulting in local 10 g SAR of 10.1 W kg−1 in an adult would deposit 2.6 W kg−1 in a neonate: an approximately fourfold difference. The SAR experienced by neonatal subjects undergoing MRI is lower than that in adults in equivalent situations. If the safety of such procedures is assessed using adult‐appropriate models then the result is a conservative estimate. © 2015 The Authors. NMR in Biomedicine published by John Wiley & Sons, Ltd.