It has previously been shown that the formation of GU base pairs in RNA copying processes leads to an accumulation of G and U in both strands of the replicating RNA, which results in a non-random distribution of base triplets. In the present paper, this distribution is calculated, and, using the X2-test, a correlation between the distribution of triplets and the amino acid composition of the evolutionarily conservative interior regions of selected globular proteins is established. It is suggested that GU wobbling in early replication of RNA could have led to the observed amino acid composition of present-day protein interiors. If this hypothesis is correct, then GU wobbling must have been very extensive in the imprecisely replicating RNA, even reaching values close to the critical for stability of its double-helical structure. Implications of the hypothesis both for the evolution of the genetic code and of proteins are discussed.