Embryonic stem (ES) cells have the capacity to give rise to all cell types of the adult organism and can be used as a model to study early cell fate decisions as they closely recapitulate in vivo events. The M and G1 phases have been suggested to be the key time windows for pluripotent cells to engage toward differentiation and SOX2 and OCT4 have been shown to orchestrate the dissolution of pluripotency and induction of cell fate commitment. Furthermore, it is known that the levels of some pluripotency transcription factors fluctuate over time, leading to the notion that ES cells experience dynamic heterogeneity, where for example low levels of NANOG can be found in cells prone to differentiation, whereas cells with high NANOG expression are in a robust pluripotent state. However, it remains unclear if and how (i) the action of transcription factors in specific cell cycle phases and (ii) protein level variations of factors that show mild differences between single cells impact cell fate decisions. This is in part due to the lack of quantitative single-cell meth-ods allowing dissecting the origin and functional consequences of gene expression heterogeneity. In this work, we established a method based on internal tagging of endogenous proteins with an excisable reporter to perform absolute molecule quantification in single living cells. This enabled us to investigate fluctuations of endogenous protein expression levels, the heterogeneity of degradation rates between individual cells and determine absolute amounts of proteins synthesized over the cell cycle. Additionally, inducible excision of the tag enabled us to estimate endogenous mRNA half-lives. Next, by using live-cell microscopy of fusion proteins, we show that SOX2 and OCT4 stay bound to mitotic chromosomes through their DNA binding domains. We then characterized the function of SOX2 at the M-G1 transition, showing that its absence specifically in this phase impairs pluripotency maintenance and abolishes its ability to induce neuroectodermal commitment. This suggests, that transcription factor activity at the M-G1 transition is essential for cell fate maintenance and differentiation. Lastly, we show that mild endogenous fluctuations in the levels of key transcription factors in ES cells bias cell fate decisions. We found that high OCT4 levels at the onset of differentiation increase the probability of single cells to commit to both neuroectoderm and mesendoderm. Further, we showed that high SOX2 levels result in a higher number of cells acquiring a neuroectodermal cell fate. This not only shows that the differentiation potential of embryonic stem cells is highly sensitive to small variations in intracellular concentrations of pluripotency transcription factors but further suggests that endogenous fluctuations of these are a major source of heterogeneity in cell fate decisions. In conclusion, the presented work identifies that the SOX2 activity in the M-G1 transition is important for cell fate maintenance and differentiation. Furthermore, the heterogeneity of protein levels governs cell fate decisions, suggesting the existence of different short-lived cell states in populations of presumed homogenous cells. We think that the minor differences in the levels of transcription factors and the cell cycle specific sensitivity of cells to transcription factor activity could be important mechanisms guiding maintenance and differentiation potential in various stem cell types.