Increasingly, mobile telephones are becoming pocket-sized and are being left in the shirt pocket with a connection to the ear for hands-free operation. We have considered an anatomic model of the chest and a planar phantom recommended by US FCC to compare the peak 1 and 10 g SARs for four typical cellular telephones, two each at 835 and 1900 MHz. An agreement within +/- 10% is obtained between calculated and experimental 1 and 10 g SARs for various separations (2-8 mm) from the planar phantom used to represent different thicknesses of the clothing both for the antenna away from or turned back towards the body. Because of the closer placement of the antennas relative to the body, the peak 1 and 10 g SARs are considerably higher (by a factor of 2-7) for pocket-mounted telephones as compared to the SARs obtained using a 6 mm thick plastic ear head model--a procedure presently accepted both in the US and Europe. This implies that a telephone tested for SAR compliance against the model of the head may be severely out of compliance if it were placed in the shirt pocket.