Viruses that replicate in the cytoplasm cannot access the host nuclear capping machinery. These viruses have evolved viral methyltransferase(s) to methylate N-7 and 2'-O cap of their RNA; alternatively, they "snatch" host mRNA cap to form the 5' end of viral RNA. The function of 2'-O methylation of viral RNA cap is to mimic cellular mRNA and to evade host innate immune restriction. A cytoplasmic virus defective in 2'-O methylation is replicative, but its viral RNA lacks 2'-O methylation and is recognized and eliminated by the host immune response. Such a mutant virus could be rationally designed as a live attenuated vaccine. Here, we use Japanese encephalitis virus (JEV), an important mosquito-borne flavivirus, to prove this novel vaccine concept. We show that JEV methyltransferase is responsible for both N-7 and 2'-O cap methylations as well as evasion of host innate immune response. Recombinant virus completely defective in 2'-O methylation was stable in cell culture after being passaged for >30 days. The mutant virus was attenuated in mice, elicited robust humoral and cellular immune responses, and retained the engineered mutation in vivo. A single dose of immunization induced full protection against lethal challenge with JEV strains in mice. Mechanistically, the attenuation phenotype was attributed to the enhanced sensitivity of the mutant virus to the antiviral effects of interferon and IFIT proteins. Collectively, the results demonstrate the feasibility of using 2'-O methylation-defective virus as a vaccine approach; this vaccine approach should be applicable to other flaviviruses and nonflaviviruses that encode their own viral 2'-O methyltransferases.