Résumé : The non-random distribution of degenerate code words in Bacteriophage MS2 RNA can be explained partially by considerations of the stability of the codon-anticodon complex in prokaryotic systems. Supporting this hypothesis we note that wobble codons are positively selected in codons having G and/or C in the first two positions. In contrast, wobble codons are statitically less likely in codons composed of A and U in the first two positions. Analyses of nucleotides adjacent to 5′ and 3′ ends of codons indicate a nonrandom distribution as well. It is thus likely that some elements of RNA evolution are independent of the structural needs of the RNA itself and of the translated protein product. © 1978 Springer-Verlag.