Selenocysteine synthesis is achieved on a specific tRNA, tRNA(Sec), which is first charged with serine to yield seryl-tRNA(Sec). Eukaryotic tRNA(Sec) exhibits an aminoacyl acceptor stem with a unique length of 9 base pairs. Within this stem, two base pairs, G5a.U67b and U6.U67, drew our attention, whose non-Watson-Crick status is maintained in the course of evolution either through U6.U67 base conservation or base covariation at G5a.U67b. Single or double point mutations were performed, which modified the identity of either or both of the base pairs. Serylation by seryl-tRNA synthetase was unaffected by substitutions at either G5a.U67b or U6.U67. Instead, and quite surprisingly, changing G5a.U67b and U6.U67 to G5a-C67b/U6.G67 or G5a-C67b/C6-G67 gave rise to a tRNA(Sec) mutant exhibiting a gain of function in serylation. This finding sheds light on the negative influence born by a few base pairs in the acceptor stem of tRNA(Sec) on its serylation abilities. The tRNA(Sec) capacities to support selenocysteylation were next examined with regard to a possible role played by the two non-Watson-Crick base pairs and the unique length of the acceptor stem. It first emerges from our study that tRNA(Sec) transcribed in vitro is able to support selenocysteylation. Second, none of the point mutations engineered at G5a.U67b and/or U6.U67 significantly modified the selenocysteylation level. In contrast, reduction of the acceptor stem length to 8 base pairs led tRNA(Sec) to lose its ability to efficiently support selenocysteylation. Thus, our study provides strong evidence that the length of the acceptor stem is of prime importance for the serine to selenocysteine conversion step.