SUMMARY: Composite manufacturing techniques are subject to a range of disturbances, due to material and processing variability. Properties of fibre reinforcements such as compaction response and permeability are very influential, as is variability in these quantities. A compaction study of four reinforcements is presented, spatial stress variations quantified using the Tekscan distributed pressure measurement system. Variability in the average stress has been explored, considering basic statistics of the spatial variation over repeated experiments, and with increasing fabric layers in the sample. Two types of behaviour have been noted, chopped strand mat and twill weave reducing the magnitude of peak stresses as additional layers are added, resulting in reduced variability in average stress between samples. The plain weave and biaxial stitched fabrics maintained peak stresses with the addition of layers, and exhibited increasing variability. The noted spatial variability has been used to specify a model permeability field for a single layer of twill weave. Random fluctuations have been imposed on this field, and 1000 permeability fields have been generated for a 4 layer preform. The upper and lower permeability bounds have been used to predict injection pressure histories for unidirectional filling, results correlating well with observed experimental variability. This study shows that if reinforcement is well known in terms of statistical variations, a simulation can be more predictive in terms of injection responses.