Objective. The International Commission on Non-Ionizing Radiation Protection guidelines and IEEE C95.1-2019 standard for human protection from local electromagnetic field exposure above 6 GHz state that absorbed (or epithelial) power density (APD) and incident power density (IPD), averaged over a square area, are internal and external physical quantities, respectively, that set the exposure limit. Per exposure standards, the measurement procedure and evaluation of the IPD have been established in technical standards, where a circular averaging area is recommended only for non-planar surfaces in IEC/IEEE 63195-1 and -2. In this study, the effects of two averaging shapes on the APD and IPD are evaluated computationally to provide new insights from the viewpoint of exposure standards. Approach. The relation between the APD, IPD, and the steady-state temperature rise (heating factor) in rectangular and human models for exposure to a single dipole, dipole arrays, and the Gaussian beams is investigated computationally with finite-difference method. Main results. The maximum differences in the heating factor of the APD and IPD for square and circular averaging areas were 4.1% and 4.4% for the antenna–model distance >5 mm, respectively. These differences appear when the beam pattern on the model surface has an elliptical shape. For an antenna–model distance ≤5 mm and at frequencies ≤15 GHz, the heating factors for square averaging areas were not always conservative to those for circular ones (−7.8% for IPD), where only the antenna feed point are visible before beam formation. Significance. The heating factors of the APD and IPD for a circular averaging area are conservative for near-field exposure of canonical sources for frequencies up to 300 GHz, except for a beam with a significant major-to-minor axis ratio and an angle of 30°–60° to a square averaging area. This tendency would help bridge the gap between exposure and product standards.