Codon usage bias is commonly studied in relation to DNA composition, while its association with the physicochemical properties of the encoded proteins has been explored to a much lesser extent. In this work, we analyze, at the taxonomic level, the relationship between the average proteome hydrophobicity, measured according to the Kyte–Doolittle scale, and the Effective Number of Codons (ENC). The analysis is based exclusively on median values derived from the aggregated statistics provided by the CompoDynamics database. The representation of the nine major biological groups in the hydrophobicity–ENC space reveals a structured and non-random distribution, with distinct regions occupied by Archaea, Bacteria, and eukaryotes. Viruses are located within the region associated with eukaryotic groups, consistently with their dependence on host translational systems. These results indicate that simple aggregated metrics are sufficient to relate the degree of codon usage degeneracy to the global physical and chemical properties of proteomes. This approach may offer a complementary perspective for the large-scale study of codon usage bias.