Elderly patients are prone to cognitive impairment and memory loss after surgical operations. This perioperative cerebral damage, named postoperative cognitive dysfunction (POCD), is profoundly affected by anesthesia. N6-methyladenosine (m6A) RNA methylation is a widely-studied epigenetic modification to regulate gene expression; however, is has never been studied in POCD. In the present study, elderly POCD mouse models were constructed using sevoflurane, and we observed a compromised global m6A RNA methylation in the mice’s hippocampuses compared with the control. Our RIP-Seq data suggested that 1244 genes (SOX2, SYN1, and BDNF) showed m6A RNA methylation in their 5′UTRs, which was significantly lower than that in the control; while only 56 genes (BACE1 and IL17A) showed m6A RNA methylation in their 5′UTRs, which was significantly higher than that in the control. Unexpectedly, m6A RNA methylation with significant differences in exons, introns, or 3′UTRs was observed in only few genes. Although we failed to find any differences in the expression of m6A-associated proteins, such as m6A “writers”, “erasers”, and “readers”, between the sevoflurane treatment and control groups, RIP-qPCR assays indicated that the binding affinity of METTL3 on mRNA 5′UTRs was particularly weakened in target genes by sevoflurane. Finally, we found that phosphorylation of METTL3 could be reduced by sevoflurane because of the inactivation of the MAPK/ERK pathway. Overall, our study determined that the inactivation of METTL3 in the mouse hippocampus, induced by sevoflurane-mediated MAPK/ERK suppression in vivo, resulted in a perturbation in m6A RNA methylation signals in the pathogenesis of POCD.