We report a cluster of four tRNA genes from Mycoplasma mycoides as well as the sequence of the alanine, proline, and valine tRNAs and the serine tRNA reading the UCN codons (where N stands for G, A, C, or U). This brings the total number of tRNA genes that we have so far characterized in this organism to 14, 6 of which code for tRNAs that read the codons of family boxes. In each of these latter cases, we found only one gene per family box, and the gene sequence contains a thymidine in the position corresponding to the wobble nucleotide, with the exception of the arginine tRNA gene that has an adenosine in this position. Furthermore, all of the tRNA structures reported here have an unsubstituted uridine in the wobble position. These rindings are similar to those reported for mitochondria, especially yeast mitochondria, that contain an arginine tRNA with the anticodon ACG. However, the resemblance is not complete since we have demonstrated the presence of two isoacceptor tRNAs for threonine having uridine and adenosine, respectively, in the wobble position. It is suggested that in the M. mycoides at least some ofthe family codon boxes are read by only one tRNA each, using an unconventional method without discrimination between the nucleotides in the third codon position.The genetic code can be thought of as being made up of sixteen boxes with four codons in each box. All codons in a box have the same nucleotides in the first two positions; the variation between the codons is in the third position. In half of the boxes all four codons specify the same amino acid, and in what follows we will refer to such boxes as family boxes (Fig. 1). Regarding the codon families it makes no difference, as far as translational fidelity is concerned, how the third codon nucleotide is read since the first two are enough to specify the amino acid. In the rest of the boxes the codons have been divided up between amino acids or between amino acids and stop words. When reading the codons of these boxes, it is obviously necessary to be able to discriminate among nucleotides in the third codon position to avoid translational errors. To explain such discrimination, Crick in his wobble rules (1) excludes a number of pairings between the third nucleotide of the codon and the wobble nucleotide of the anticodon on structural grounds.Most of the accumulated data concerning codon reading in vivo indicate that codons are read in the cell according to the wobble rules. Nevertheless, there is evidence in the literature for codon readings in vivo involving almost all of the forbidden base pairs. These exceptions to the rules are oftwo different types, (i) unconventional reading with discrimination among the nucleotides in the third codon position, albeit using principles other than those laid down in the wobble rules, and (ii) reading without discrimination in this position