It has been proposed that patterns in the usage of synonymous codons provide evidence that individual tRNA molecules are recycled through the ribosome, translating several occurrences of the same amino acid before diffusing away. The claimed evidence is based on counting the frequency with which pairs of synonymous codons are used at nearby occurrences of the same amino acid, as compared to the frequency expected if each codon were chosen independently from a single genome-wide distribution. We show that such statistics simply measure variation in codon preferences across a genome. As a negative control on the potential contribution of pressure to exploit tRNA recycling on these signals, we examine correlations in the usage of codons that encode different amino acids. We find that these controls are statistically as strong as the claimed evidence and conclude that there is no informatic evidence that tRNA recycling is a force shaping codon usage. Display omitted •Synonymous codon usage patterns are known to vary across the genome and within genes•Mathematically, this variation implies a diagonal-positive local covariance signal•That signal is thus not evidence for molecular tRNA reuse by the ribosome•Rather, it reflects a complicated covariance structure across 61 codons Hussmann and Press find that observed excesses in pairs of coding sequence occurrences of the same synonymous codon are a generic consequence of the existence of spatial variation in codon preferences and therefore cannot be interpreted as evidence for the recycling of individual tRNA molecules during translation.