The thermoluminescent dosimeter (TLD) method has been proposed as a useful tool for estimating internal radiation absorbed dose in nuclear medicine. An efficient approach to verify the accuracy of the TLD method has been performed in this study. Under the standard protocol for 2-[F-18]fluoro-2-deoxy-D-glucose (18F-FDG), whole body PET experiments and simultaneous body surface dose measurements by TLDs were performed on six normal volunteers. By using the body surface dose measured with TLDs, the cumulated activities of nine source organs were estimated with a mathematical unfolding technique for three different initial guesses. The accuracy of the results obtained by the TLD method was investigated by comparison with the actual cumulated activity of the same source organs measured by whole body PET. The cumulated activities of the source organs obtained by the TLD method and whole body PET show a significant correlation (correlation coefficient, r > 0.98, level of confidence, p < 0.001) with each other. The mean effective doses in this study are 3.2 x 10(-2) mSv MBq(-1) obtained from the TLD method and 2.9 x 10(-2) mSv MBq(-1) obtained from the whole body PET. Good agreement between the results of the TLD method and whole body PET was observed.