Depth profiles of nitrogen implanted into Zr with an energy of 50 keV were calculated by dynamic SASAMAL code with three different assumptions for the diffusion of excess atoms over stoichiometry, i.e., ‘no diffusion', ‘both-sides-diffusion’ and ‘upward-diffusion'. To distinguish nitrogens implanted certain stage of implantations, alternate implantations of15N and14N were used. The results were compared with the experimental results by the resonance nuclear reaction analysis, NRA. For15N implantation with fluences from 1 × 1017to 1 × 1018ions/cm2, the calculated results with ‘upward-diffusion’ agreed very well with the NRA results for all fluences. For the depth profile of pre-implanted15N (1 × 1017ions/cm2), which was changed by the subsequent14N implantation with fluences of 1 ∼ 10 × 1017ions/cm2, the agreement with the NRA results was satisfactory until the14N fluence did not exceed 5 × 1017ions/cm2, but for higher fluences, the retained probabilities of15N obtained by the ‘upward-diffusion’ code were too low compared with the experimental value obtained by NRA. For the depth profiles of15N (1 × 1017ions/cm2) implanted following after implantations of14N with fluences of 1 ∼ 10 × 1017ions/cm2, the agreement with the NRA results was quite good for all14N fluences. It is concluded that the approximation of ‘upward-diffution’ is proper satisfactorily for the treatment of atoms implanted at the final stage of implantations, but a problem is left for the treatment of atoms implanted at the early stage of implantations.