Casting metal alloys in the semi-solid state is now becoming a well established manufacturing technique. But, the success of this technology necessitates a good understanding of the feedstock material behaviour. To obtain high quality components with semi-solid metal processing, a homogeneous distribution of phases must be maintained in the material during the die filling stage. Many parameters affect the process such as temperature, time and stress history, which influence the shape, size and connectivity of the particles that make up the slurry. The subsequent phase interaction mechanisms are quite complex and have direct effects on the flow and final micro-structure distribution of the cast part and thus, without any doubt, on its mechanical properties. Two-phase numerical models have been developed to account for the liquid-solid phase separation e.g. [1,2]. Several two-phase models have been elaborated on the basis of soil mechanics and consider that the phase interaction term is mainly due to the flow through a porous medium. Because of the difficulties of making direct measurements in an extremely hostile environment, there has been very little work done to validate these models. In order to fill this gap, a better understanding of the phase distribution and phase segregation mechanisms during the filling step is required. In this work, the post-solidification primary α-phase distribution inside an industrial semi-solid cast part has thus been investigated. A thorough metallographic analysis has been performed using an upright microscope coupled to a Clemex image-analysis software. The results were then processed to produce a map of the final α-phase distribution. Many different grain scales have been observed in the solidified part and their distributions seem to be closely associated to the velocity field. Contacts between moving particles seem to play an important role in the phase distribution and show many similarities to granular materials. This latter aspect should be considered in the development of new constitutive models for semi-solid slurries.