The existence of a strong stochastic component in gene expression have now been clearly demonstrated: despite constant environmental conditions, genetically identical cells do show significant fluctuation in their gene expression levels. Such a stochastic behaviour has been involved in an ever growing amount of biological processes (for a recent review, see Eldar and Elowitz, 2010). Among various possible sources of stochasticity in gene expression (SGE), several studies suggested that the chromatin environment could be involved in regulating the level of stochasticity in gene expression. In order to investigate this relationship in higher eukaryotic cells, we have transfected 6C2 chicken erythrocyte progenitors cells to express mCherry or hKo reporter fluorescent proteins under the control of the CMV promoter. The transfection conditions we used have shown, in previous experiments, the existence in most of the stably transfected cells of one single genomic insertion site for the transgene. We selected cells demonstrating a stable integration of the transgene and, by sorting individual cells, we generated clones expressing the reporter genes. FACS analysis of those clones showed the expected clone-to-clone differences in gene expression mean. More importantly, we also detected a substantial amount of clone-to-clone differences in the variation around the mean, by using normalized variance (σ2/µ2), a robust stochastic indicator, which is based on the variance estimation but also takes into account the mean level of gene expression. For each clone, we have the full distribution of expression levels, which serves as a basis for modeling the molecular mechanisms involved We started first by a simple model in which transcription is initiated with a certain probability. This model generates a distribution with a specific signature for two consecutive Poisson processes. However, since our data do not exhibit such a signature, a more elaborate model was used: the two-state model in which the promoter can be either open or closed, and has a certain probability to initiate transcription when in the open state (Paulsson, 2005). For this model, we used an analytical formula relating the mean value of gene expression with its normalized variance to interpret our data. This analytical formula is limited in two ways: (1) it does not take into account the full distribution of expression levels, and (2) it can not be used when the distribution is not at equilibrium. We can, for example, not use it to model the state of a system having suffered a disturbance that shifts it away from its equilibrium, which is the case with treatment by global chromatin modifiers on a scale of short time. To model such events we will use the algorithm of Gillespie (1976), to simulate stochastic biochemical events. This model should then allow us to make experimentally testable predictions, such as: what will the impact of inhibitors of various epigenetic modifications, such as those induced by the addition of modifying the overall state of chromatin as TSA or AZA-C on the SEG? These predictions can then be compared with experimental evidence obtained in these treatments. Eldar, A., and Elowitz, M.B. (2010). Functional roles for noise in genetic circuits. Nature 467, 167-173. Gillespie, D.T. (1976). A General Method for Numerically Simulating the Stochastic Time Paulsson, J. (2005). Models of stochastic gene expression. Phys Life Rev 2, 157-175.