In the present research, by the use of the MCNP computational code that is based and set on the Monte Carlo method, at first, pulse height distribution for the known sources, Cs-137and Co-60, with the F8 tally has been obtained for the NaI(Tl) detector. Then the pulse height distribution has been obtained experimentally for these two sources. After the comparison of simulation results with the experimental results and validation of written codes, the main codes have been written. The cylindrical NaI(Tl) detector with 3 inches length and 1.5 inches radius has been assumed. Gamma radiations with energy 0.2, 1, 5, 10 and 50 MeV have been also used and an isotropic point source has been placed in the distance ( d ) on the x -axis. The x -axis has been assumed the symmetry axis. The program has been run for 20 distances ( d ) from 0.00381 up to 381 cm. Then the intrinsic efficiency variations versus source-detector distance has been calculated for different gamma ray energies and the intrinsic efficiency variations of aI(Tl) detector versus source-detector distance have been analyzed for different gamma radiations. It should be noted, the intrinsic efficiency depends not only on photon energy but also on source-detector geometric configuration. The intrinsic efficiency variations have been analyzed by the Dirac theory and mean chord length of photons in the detector and mean interaction length at each energy. Finally, all of the previous calculations have been done for a CsI(Tl) detector with the same sources and geometric configuration. Then the CsI(Tl) detector results have been compared with those of the NaI(Tl) detector. The results showed that the variation of intrinsic efficiency graph of the both detectors have a minimum at d/R=0.7. Keywords: NaI(Tl) detectors, intrinsic efficiency, MCNP, Monte Carlo, tally, gamma radiation, isotropic source, mean chord length, mean interaction length