Abstract: Nuclear resonance fluorescence (NRF) is an emerging nondestructive active inspection technology. Specific isotopes can be identified by analyzing the characteristic energy recorded in the gamma spectrum. This technology plays an important role in the scanning inspection of small containers containing unknown explosives. In this paper, an NRF backscatter inspection scheme is optimized using Monte Carlo simulation program based on the X-ray source generated by electron accelerator bremsstrahlung radiation. After optimized design, X-ray leakage rate is reduced by 386 times, the beam asymmetry is less than 2%, the beam uniformity is more than 70%. The NRF characteristic energy spectrum of graphite samples and ammonium nitrate sample obtained by HPGe detector verifies the feasibility of the design scheme. The innovative application of importance sampling technique improves the simulation calculation efficiency of graphite sample by 72.23 times. The design scheme and calculation results provide technical support for the research and development of NRF based non-destructive nuclear detection system.