Simulation of Resin Transfer Moulding (RTM) is a key design tool for improving product quality and reducing high process development costs. Darcy's law is currently used in RTM simulation packages to model the mould filling stage on a macro-scale (length scale of a part). However, some of the assumptions of Darcy's law are not satisfied during the manufacturing process, e.g. 2-D Newtonian fluid flow without surface tension and no velocity gradients at walls or across permeability variations. This thesis presents a new multi-scale modelling approach that overcomes the limitations of Darcy's law by performing micro-scale simulations, which resolve the fluid flow on the length scale of an idealised fibre bundle. The interaction of the fluid on the fibre bundle appears as a source term (micro-model) in the momentum equations. These equations are then solved on a macro-scale using the Volume of Fluid method to simulate the mould filling stage of RTM. The micro- and macro-scale simulations are implemented within the computational fluid dynamics software CFX4 using Fortran subroutines. Laboratory scale experiments are used to determine the in-plane permeability constants for a unidirectional fabric using the parallel and radial-flow permeability methods. The techniques used for interpreting the radial-flow measurements are then compared and evaluated. The resulting constants are used in the development of a micro-scale model and the validation of the multi-scale simulations. The mould filling behaviour of thick composite structures (30mm) is investigated on a semi-industrial scale in three dimensions, using a novel sensing technology called SMARTWeave®. The measurement results highlight the influence of race-tracking on mould fill-time and flow-front shape across permeability variations. The implementation of the multi-scale model is validated against a series of new analytical solutions for a variety of flow problems. The results obtained from the multiscale model are shown to be more accurate than Darcy's law at predicting the velocity profile during mould filling, mould filling with race-tracking and also in cases when there is significant permeability variation (local and global) between fabric layers. A preliminary investigation also shows the suitability of this approach to model surface tension effects, which is of importance for low pressure injections.