This paper presents new definitions for obtaining reproducible results in numerical phone dosimetry. Numerous numerical dosimetric studies have been published about the exposure of mobile phone users which concluded with conflicting results. However, many of these studies lack reproducibility due to shortcomings in the description of the phone positioning. The new approach was tested by two groups applying two different numerical program packages to compare the specific anthropomorphic mannequin (SAM) to 14 anatomically correct head models. A novel definition for the positioning of mobile phones next to anatomically correct head models is given along with other essential parameters to be reported. The definition is solely based on anatomical characteristics of the head. A simple up-to-date phone model was used to determine the peak spatial specific absorption rate (SAR) of mobile phones in SAM and in the anatomically correct head models. The results were validated by measurements. The study clearly shows that SAM gives a conservative estimate of the exposure in anatomically correct head models for head only tissue. Depending on frequency, phone position and head size the numerically calculated 10 g averaged SAR in the pinna can be up to 2.1 times greater than the peak spatial SAR in SAM. Measurements in small structures, such as the pinna, will significantly increase the uncertainty; therefore SAM was designed for SAR assessment in the head only. Whether SAM will provide a conservative value for the pinna depends on the pinna SAR limit of the safety standard considered.