Microtubules are non-covalent polymers important for many processes such as vesicle trafficking and establishment of cell polarity and essential for cell division. A key property of microtubules is that they are highly dynamic. Microtubules constantly switch between phases of growth and shortening. This behavior is called dynamic instability. Although the importance of dynamic instability is well established from a cell-biological point of view, its regulation and mechanistic details are poorly understood. This doctoral thesis summarizes an effort to better understand dynamic instability by means of an interdisciplinary approach. The opening chapter presents an overview of microtubule dynamics and its pending questions. Then, an introduction to stochastic modeling in biology is presented for readers of diverse background. The dynamics of the microtubule network is then studied with a model at a mesoscopic scale (coarse grain) and a model at a microscopic scale (fine details). The mesoscopic modeling results indicate that many behaviors thought to require regulatory proteins are instead unavoidable outcomes of the physical constraints on a system of nucleated polymers competing in a confined space. This suggests that regulatory proteins tune microtubule dynamics rather than govern it. This conclusion has important evolutionary implications as microtubules are present in all eukaryotes and therefore their underlying mechanistic principles must be robust. The microscopic model is the first one built at a the dimer scale to recapitulate dynamic instability. With this model, the mechanistic details of the once paradoxical microtubule dilution experiments are shown. As opposed to the canonical view, the microscopic model indicates that interprotofilament cracks are always present, even when the microtubule is growing, and it also indicates that there is GTP-tubulin binding to the shortening microtubule. Quantitative analysis concludes that it is the relationship between the lengths of cracks and the GTP cap what dictates microtubule dynamics, not the GTP cap alone. With its simulation speed and level of detail, the microscopic scale model finally opens the door to testing hypotheses of the mechanisms used by microtubule regulatory proteins.