The aim of this work is to develop a numerical model capable of predicting the grain density in the Mg-based matrix phase of an AZ91/SiC composite, as a function of the total mass fraction of the embedded SiC particles. Based on earlier work in a range of alloy systems, we assume an exponential relationship between the grain density and the maximum supercooling during solidification. Analysis of data from cast samples with different thicknesses, and mass fractions of added SiCp, permits conclusions to be drawn on the role of SiCp in increasing grain density. By fitting the data, an empirical nucleation law is derived that can be used in a micro model. Numerical simulation based on the model can predict the grain density of magnesium alloys containing SiC particles, using the mass fraction of the particles as inputs. These predictions are compared with measured data.