The paper by Swain, Elowitz, and Siggia explores the stochastic nature of gene expression, which is characterized by both intrinsic and extrinsic noise. Intrinsic noise arises from the inherent randomness in biochemical processes such as transcription and translation, while extrinsic noise comes from fluctuations in other cellular components that indirectly affect gene expression. The authors provide a theoretical framework to decompose the total variation in gene expression into its intrinsic and extrinsic components. They demonstrate that simultaneous measurement of two identical genes per cell can help distinguish between these two types of noise. Analytical expressions for intrinsic noise are derived for a model that includes all major steps in transcription and translation, showing that transcription dominates when the average number of proteins produced per mRNA transcript is greater than about two. The model also accounts for gene replication and cell division, which cause protein numbers to tend to a limit cycle. The authors calculate a general form for extrinsic noise and illustrate it with the example of a single fluctuating extrinsic variable, such as a repressor protein acting on the gene of interest. The results are confirmed by stochastic simulations using parameters for Escherichia coli. The paper concludes by discussing the implications of these findings for understanding and potentially regulating gene expression noise in living cells.The paper by Swain, Elowitz, and Siggia explores the stochastic nature of gene expression, which is characterized by both intrinsic and extrinsic noise. Intrinsic noise arises from the inherent randomness in biochemical processes such as transcription and translation, while extrinsic noise comes from fluctuations in other cellular components that indirectly affect gene expression. The authors provide a theoretical framework to decompose the total variation in gene expression into its intrinsic and extrinsic components. They demonstrate that simultaneous measurement of two identical genes per cell can help distinguish between these two types of noise. Analytical expressions for intrinsic noise are derived for a model that includes all major steps in transcription and translation, showing that transcription dominates when the average number of proteins produced per mRNA transcript is greater than about two. The model also accounts for gene replication and cell division, which cause protein numbers to tend to a limit cycle. The authors calculate a general form for extrinsic noise and illustrate it with the example of a single fluctuating extrinsic variable, such as a repressor protein acting on the gene of interest. The results are confirmed by stochastic simulations using parameters for Escherichia coli. The paper concludes by discussing the implications of these findings for understanding and potentially regulating gene expression noise in living cells.