The 3-methylcytidine (m3C) modification is widely found in eukaryotic species of tRNASer, tRNAThr, and tRNAArg; at residue 32 in the anti-codon loop; and at residue e2 in the variable stem of tRNASer. Little is known about the function of this modification or about the specificity of the corresponding methyltransferase, since the gene has not been identified. We have used a primer extension assay to screen a battery of methyltransferase candidate knockout strains in the yeast Saccharomyces cerevisiae, and find that tRNAThr(IGU) from abp140-Δ strains lacks m3C. Curiously, Abp140p is composed of a poorly conserved N-terminal ORF fused by a programed +1 frameshift in budding yeasts to a C-terminal ORF containing an S-adenosylmethionine (SAM) domain that is highly conserved among eukaryotes. We show that ABP140 is required for m3C modification of substrate tRNAs, since primer extension is similarly affected for all tRNA species expected to have m3C and since quantitative analysis shows explicitly that tRNAThr(IGU) from an abp140-Δ strain lacks m3C. We also show that Abp140p (now named Trm140p) purified after expression in yeast or Escherichia coli has m3C methyltransferase activity, which is specific for tRNAThr(IGU) and not tRNAPhe and occurs specifically at C32. We suggest that the C-terminal ORF of Trm140p is necessary and sufficient for activity in vivo and in vitro, based on analysis of constructs deleted for most or all of the N-terminal ORF. We also suggest that m3C has a role in translation, since trm140-Δ trm1-Δ strains (also lacking m2,2G26) are sensitive to low concentrations of cycloheximide.