CH /N separation is one of the great challenges in gas separation, which is of scientific and practical importance, such as in the upgrading of unconventional natural gas. Unfortunately, the separation performance is still quite low so far mainly due to their very close physical properties. In this work, a high-throughput computational screening method was performed to develop metal-organic frameworks (MOFs) for efficient CH /N separation. General designing rules as well as the correlation between selectivity and our proposed adsorbility (AD) parameter were obtained by carrying out systematic GCMC simulations of the existing 5109 CoRE MOFs. With the aid of this information, five virtual MOFs were screened out from the large database with 303 991 generated MOFs constructed in our previous work, exhibiting much higher selectivities than all the reported values. Among them, the selectivity of Zn-PYZ-BPY-1 can reach over 29.0, about 2.4 times of the highest value reported in the literature. These results may not only suggest promising candidates for CH /N separation but also provide useful information for large screening of MOFs for other specific separation mixtures.