The radioactive gas Rn in indoor air shows severe harm to the human body and even induces the occurence of lung cancer. Finding an excellent candidate that can efficiently and selectively adsorb Rn, is still a great challenge. In this work, we use the grand canonical ensemble Monte Carlo simulation to systematically investigate the selectivities of 23 different kinds of representative metal organic framework (MOFs) for Rn separation from the Rn/N-2 and Rn/O-2 mixtures at 298 K. Results indicate that ZIF-12, HKUST-1, IRMOF-62 and ZIF-11 are four kinds of excellent candidates for capturing Rn. In particular, the selectivities of ZIF-12 for Rn/N-2 and Rn/O-2 mixtures with Rn molar fraction X-Rn = 0.001 reach similar to 2500 and 1200 at 1 atm and 298 K, respectively. Moreover, when the Rn concentration becomes smaller, the selectivity of ZIF-12 for Rn would increase significantly. In short, we found that ZIF-12 is indeed a very promising candidate for trapping radioactive Rn in indoor air.