As coronaviruses (CoVs) replicate in the host cell cytoplasm, they rely on their own capping machinery to ensure the efficient translation of their mRNAs, protect them from degradation by cellular 5 exoribonucleases, and escape innate immune sensing. The CoV nonstructural protein 14 (nsp14) is a bi-functional replicase subunit harboring an N-terminal 3-to-5exoribonuclease (ExoN) domain and a C-terminal (N7-guanine)-methyltransferase (N7-MTase) domain that is assumed to be involved in viral mRNA capping. Here, we first revisited the crystal structure of severe acute respiratory syndrome (SARS)-CoV nsp14 to perform an in silico comparative analysis between different betacoronaviruses (beta-CoVs). In this study, we identified several residues likely to be involved in the formation of the catalytic pocket of N7MTase, which presents a fold that is distinct from the Rossmann fold observed in most known MTases. Next, for multiple beta-CoVs, site-directed mutagenesis of selected residues was used to assess their importance for in vitro enzymatic activity and viral replication in cell culture. For SARS-CoV and Middle East respiratory syndrome-CoV, most of the engineered mutations abolished the N7-MTase function, while not affecting nsp14-ExoN activity. Upon reverse engineering of these mutations into beta-CoV genomes, we identified two substitutions (R310A and F426A in SARS-CoV) that abrogated viral progeny production and one mutation (H424A) that yielded a crippled phenotype across all beta-CoVs tested. Our results identify the N7-MTase as a critical enzyme for beta-CoV replication and defined key residues of its catalytic pocket that can be targeted to design inhibitors with a potential pan-coronaviral activity spectrum.