Pyrrolysine and selenocysteine have infiltrated natural genetic codes via the translation of canonical stop codons. UGA translation as selenocysteine is absolutely dependent on message context. Here we describe the first experimental examination of contextual requirements for UAG translation as pyrrolysine. A hexahistidine-tagged Methanosarcina barkeri mtmB1 gene, encoding monomethylamine methyltransferase MtmB1, was introduced into Methanosarcina acetivorans. Host mtmB expression was minimized by growth on methanol and recombinant mtmB1 products monitored by anti-MtmB and anti-hexahistidine immunoblotting. UAG translation was not compromised, as recombinant MtmB1 was 1% of cellular protein with only trace UAG-terminated mtmB1 product detectable. Untranslated regions flanking mtmB1 were not required for UAG translation, but loss of a downstream pyrrolysine insertion sequence (PYLIS) significantly increased the UAG-termination product of mtmB1 and decreased the UAG-translation product, which nonetheless contained pyrrolysine. An in-frame UAG within a bacterial uidA transcript was translated in the methanogen as pyrrolysine with 20% efficiency, suggesting UAG translation in the absence of evolved context. However, predominant UAG-directed termination with enhancement of UAG translation by the PYLIS appears analogous to cis-acting elements for UGA translation as selenocysteine, although different mechanisms may underlie these recoding events.