Mathematical modelling is increasingly being used to generate testable models of complex biological processes and predict how cells behave under particular conditions. On p. 1381, Alex Mogilner and co-workers provide a good example of this, introducing a mathematical model that describes microtubule self-organization in fish melanocyte fragments. In these fragments, dynein motors associated with pigment granules are able to direct formation of radial microtubule arrays in the absence of centrosomes. Mogilner and co-workers base their model on the observed sequence of events that follow stimulation of these motors: granule clustering, nucleation of microtubules, and further clustering to produce a single microtubule array. In addition, they incorporate numerous experimentally determined parameters, such as the microtubule treadmilling rate, the size of the melanocyte fragments and the granule dissociation rate. In computational simulations, the model accurately reproduces the pathways of granule aggregation and microtubule self-organization observed in vivo. The authors anticipate that it will not only yield further insights into the intrinsic properties of self-organizing microtubule/motor systems but also shed light on centrosome-governed microtubule arrays.