Determining and understanding the multiple scaled phenomena that control fluid flow through a fiber preform is a prerequisite for developing resin transfer molding (RTM) as a viable process for manufacturing complex composite structures. In this paper we use computational and experimental simulations to study macroscopic scale filling of a 50% fiber loaded cubic shell. The computational simulation assumes that the thickness of the mold is small relative to the local plane of the flow and that the flow is governed by the well known Darcy Law. Experiments were first performed in transparent flat plate molds to measure the permeability of the cloth preforms. A transparent cubic shell mold was constructed to compare the experimental fill behavior with the computer simulation predictions for a more realistic structure. All the experiments presented here used Mobil DTE 24 lubricating oil, viscosity = 80 cP and were performed at room temperature with 20 psig injection pressure and atmospheric pressure at the mold outlet. Mold design allowed maximum visibility of the flow field during experiments. A video camera system was used to record the fill experiments for data acquisition and for comparison to model predictions. Experimentally determined fill rates are dependent upon the particular configuration of the cloth lay-up. The computer simulations were used to examine the macroscopic model consequences of the individual lay-ups within the range of the experiments. The ability of the numerical simulation to predict mold filling behavior in cases where realistic preform edge defects exist was evaluated. The results indicate that qualitative agreement exists between the computed and observed flow fronts and fill times. It was found that the code used was capable of simulating flowfields in conditions where edgeflow behavior was enhanced or restricted. Such studies provide insight into the effect of tool and component design decisions on the filling behavior of complex components.