Gene expression noise is a universal phenomenon across all life forms. Although beneficial under certain circumstances, expression noise is generally thought to be deleterious. However, neither the magnitude of the deleterious effect nor the primary mechanism of this effect is known. Here, we model the impact of expression noise on the fitness of unicellular organisms by considering the influence of suboptimal expressions of enzymes on the rate of biomass production and the energetic cost associated with imprecise amounts of protein synthesis. Our theoretical modeling and empirical analysis of yeast data show four findings. (i) Expression noise reduces the mean fitness of a cell by at least 25%, and this reduction cannot be substantially alleviated by gene overexpression. (ii) Higher sensitivity of fitness to the expression fluctuations of essential genes than nonessential genes creates stronger selection against noise in essential genes, resulting in a decrease in their noise. (iii) Reduction of expression noise by genome doubling offers a substantial fitness advantage to diploids over haploids, even in the absence of sex. (iv) Expression noise generates fitness variation among isogenic cells, which lowers the efficacy of natural selection similar to the effect of population shrinkage. Thus, expression noise renders organisms both less adapted and less adaptable. Because expression noise is only one of many manifestations of the stochasticity in cellular molecular processes, our results suggest a much more fundamental role of molecular stochasticity in evolution than is currently appreciated.flux balance analysis | metabolic network M any cellular processes are subject to substantial stochastic variation, because they depend on interactions among a small number of molecules. For example, the transcriptional initiation of a gene in a haploid cell relies on the binding of the transcriptional machinery to a single DNA molecule. The stochastic nature of this and other molecular interactions results in a large variation in the expression of the same gene by different isogenic cells under the same environment (1-3). How and to what extent such molecule-level stochasticity affects the wellbeing of organisms and their evolution is not well-understood. Recent experimental characterization of gene expression noise in multiple species (4-9) provides necessary empirical data for addressing these fundamental but wide-open questions.In this work, gene expression noise refers to the stochastic variation in the protein expression level of a gene among isogenic cells in a homogenous environment (10). Expression noise arises from both intrinsic and extrinsic variations (2,6,8,10,11). Stochastic events in gene expression, including those in transcriptional initiation, mRNA degradation, translational initiation, and protein degradation, generate intrinsic noise (10). Differences among cells, either in local environment or the concentration or activity of any factor influencing gene expression, generate extrinsic noise (10). Extrin...