To ensure that the radio-frequency (RF) energy absorbed by human subjects during MRI does not produce any harmful health effects to the patient - including local thermal damage or whole-body thermoregulatory challenges - regulatory entities have set limits on the maximum local, partial-body, and whole-body specific absorption rates (SAR) of RF energy. This paper discusses numerical calculations of partial-body and local peak specific absorption rate. These are compared to currently used partial-body average SAR calculations in specifying the partial-body average SAR limits and local peak SAR that are based on body mass. Calculations using the fi nite-difference time-domain numerical method and 2 mm resolution Visible Human models showed that while there were some differences in the two magnetic field strengths (1.5 T and 3.0 T or 64 MHz and 128 MHz), the gross patterns of SAR distribution were similar for both 1.5 T and 3.0 T birdcage MRI coils. Computed partial-body SARs were higher than those from curve-fitting formulas used for male and female local peak SARs (in 1 g or 10 g averaging regions), and were considerably greater than specified in the regulatory limits for both whole-body and partial body SARs. In all the cases studied, with data both for 1.5 T and 3.0 T birdcage MRI coils, computed partial-body SARs exceeded values given by regulatory limits for the patient. The local peak SARs may exceed values specified in IEC and FDA regulatory limits for both whole-body and partial-body SARs. This was because the mass-ratio-based calculations are not directly related to SARs in specific body tissues.