Centrosomes are the major microtubule organizing centers of animal cells. The two centrosomes present at the onset of mitosis must separate in a timely fashion along the nuclear envelope to ensure proper bipolar spindle assembly and thus genome stability. Microtubuleassociated motors of the kinesin-5 family are required for centrosome separation in several systems, but are partially redundant or entirely dispensable in others, where the minus-end directed motor dynein plays an important role. The mechanisms by which dynein powers centrosome separation are incompletely understood. Furthermore, the nature of the symmetry-breaking mechanisms that imbalance the forces acting on centrosomes to favor theirmovement away from each other are not known. We addressed these questions using a combination of 3D time-lapse microscopy, image processing and computational modeling to dissect centrosome separation in the polarized one-cell C. elegans embryo that entirely relies on dynein for this process. First, we have characterized the quantitative features of centrosome separation in the wild-type. Next, we compared centrosome separation between wild-type and mutant/RNAi conditions. Our analysis revealed that centrosome separation is powered by the combined action of dynein at the nuclear envelope and at the cell cortex. Moreover, we demonstrated that cortical dynein requires actomyosin contractility to separate centrosomes. These observations suggest that cortical dynein acts by harnessing anterior-directed actomyosin cortical flows initiated earlier in the cell cycle by the centrosomes themselves. To confirmthismodel, we successfully tested experimentally two of its key predictions, namely that dynein complexes flow toward the anterior together with the cortex and that the velocity of centrosome separation correlates with that of the flow of the nearby cortex. Taken together, these results demonstrate that centrosome separation is driven by nuclear and cortical dynein, where the latter acts by transmitting forces produced by the cortical actomyosin flow. To test whether this model is sufficient to explain centrosome separation, we developed a 3D computationalmodel of cytoskeleton dynamics. Indeed, predicted centrosome separation agrees quantitatively with the experimental observations in wild-type and mutant/RNAi conditions. Moreover, the qualitative predictions of the model are robust for parameter changes. Furthermore, computational simulations demonstrate that forces are intrinsically organized to move centrosomes away from each other without the need of any extrinsic symmetry-breaking mechanism. Indeed, in the case of nuclear dynein-driven separation, the position of centrosomes between the nuclear envelope and the cortex results in an asymmetric microtubule aster that leads to centrosome outward movement. In the case of cortical dynein, cortical flows are triggered by centrosomes and always move away from them, such that their forces are always directed to separate centrosomes. Therefore, this separation mechanism functions irrespective of the initial position of centrosomes along the cortex. In conclusion, in this thesis we uncover a novel organizing principle in which dynein, coupled with cell geometry and flow pattern, serves to robustly separate centrosomes and thus ensure genome stability.