Vacuum-assisted resin transfer molding (VARTM) is a manufacturing process that is used to make large and complex composite structures. While promising, VARTM still suffers from relatively low fiber volume fractions and high void content in the final products. The infusion step of VARTM is very important, because the quality of the final product is usually decided by this process. Consequently, a comprehensive understanding of the infusion process is essential. In this study, a three-dimensional digital image correlation (3D-DIC) testing system was set up to research the entire infusion process through the monitor of the thickness change of the laminates in this process. Two distinct VARTM processes, with and without a rigid cover mold, were designed to be studied. The 3D-DIC technique proved to be a valid method that not only can monitor the thickness evolution of isolated points but also can give a full-field distribution of the thickness change of the laminate. The results showed that, without the use of a rigid cover mold, the stack of reinforcements initially shrank and then expanded as the resin filled the cavities before closing the inlet, while when using a rigid cover mold there was an additional expansion period before the shrinkage occurred. Such an expansion stage could promote the flow of the resin, shortening the infusion time.