ABSTRACT Redundancy in the genetic code allows for differences in transcription and/or translation efficiency between sequences carrying synonymous polymorphisms, potentially leading to phenotypic changes. It is commonly admitted that the evolution codon usage bias (“CUB”, the over-representation of certain codons in a genome, a gene or in positions along a gene) are driven by a combination of neutral and selective processes, but their relative contribution is a matter of debate, especially in mammals. Particularly, integrative studies quantifying the phenotypic impact of CUB at different molecular and cellular levels are lacking. Here we report a multiscale analysis of the effects of synonymous codon recoding during heterologous gene expression in human cells. Six synonymous versions of the shble antibiotic resistance gene were generated, fused to a fluorescent reporter, and independently expressed in HEK293 cells. Multiscale phenotype was assessed by means of: i) mRNA-to-DNA and protein-to-mRNA ratios for each shble version; ii) cellular fluorescence, using flow cytometry, as a proxy for single-cell level expression; and iii) real-time cell proliferation in absence or presence of antibiotic, as a proxy for the cell fitness. We show that differences in CUB strongly impact the molecular and the cellular phenotype: i) they result in large differences in mRNA and in protein levels, as well in mRNA-to-protein ratio; ii) they introduce splicing events not predicted by current algorithms; iii) they lead to reproducible phenotypic heterogeneity; iv) they lead to a trade-off between the benefit of antibiotic resistance and the burden of heterologous expression. We interpret that CUB modulate mRNA availability and suitability for translation in human cells, leading to differences in protein levels and eventually eliciting phenotypic differences. AUTHOR SUMMARY The genetic code is redundant, with several codons encoding for the same amino acid. These synonymous codons are not used with equal frequencies. Instead, codon usage bias (CUB) varies between species, genes, and even positions along a gene. At each of these levels, CUB are shaped by the overall balance between mutational biases and selection forces. To shed light on the molecular mechanisms underlying molecular and organism phenotypes, integrative studies quantifying the phenotypic impact of CUB at different levels of biological integration are necessary. Here, we monitored the multiscale changes induced by six synonymous versions of an antibiotic resistance gene independently expressed in a human cell line. We show that: 1. both mRNA levels, protein levels, and their ratios are affected by CUB; 2. potential effects on mRNA (splicing and availability for translation) seem to constitute the main level of action; 3. cell fitness is severely impacted by a trade-off between the burden of heterologous expression and the benefits of antibiotic resistance. This integrative study provides new insights on translation regulation and the associated phenotypic impact in human cells, associated to CU.