Transfer RNAs of the extreme halophile Haloferax volcanii contain several modified nucleosides, among them 1-methylpseudouridine (m1 psi), pseudouridine (psi), 2'-0-methylcytosine (Cm) and 1-methylinosine (m1l), present in positions 54, 55, 56 and 57 of the psi-loop, respectively. At the same positions in tRNAs from eubacteria and eukaryotes, ribothymidine (T-54), pseudouridine (psi-55), non-modified cytosine (C-56) and non-modified adenosine or guanosine (A-57 or G-57) are found in the so-called T psi-loop. Using as substrate a T7 transcript of Haloferax volcanii tRNA(Ile) devoid of modified nucleosides, the enzymatic activities of several tRNA modification enzymes, including those for m1 psi-54, psi-55, Cm-56 and m1l-57, were detected in cell extracts of H.volcanii. Here, we demonstrate that modification of A-57 into m1l-57 in H.volcanii tRNA(Ile) occurs via a two-step enzymatic process. The first step corresponds to the formation of m1A-57 catalyzed by a S-adenosylmethionine-dependent tRNA methyltransferase, followed by the deamination of the 6-amino group of the adenine moiety by a 1-methyladenosine-57 deaminase. This enzymatic pathway differs from that leading to the formation of m1l-37 in the anticodon loop of eukaryotic tRNA(Ala). In the latter case, inosine-37 formation preceeds the S-adenosylmethionine-dependent methylation of l-37 into m1l-37. Thus, enzymatic strategies for catalyzing the formation of 1-methylinosine in tRNAs differ in organisms from distinct evolutionary kingdoms.