Woven composites are widely utilised across diverse industrial sectors. In recent de\-cades, considerable research has been devoted to determine their properties. A key advantage of woven composites lies in their ability to absorb a significant amount of energy during impact events. However, the absence of a precise failure criterion presents a challenge in accurately predicting their failure behaviour, thereby hindering the accurate assessment of the crashworthiness of large structures. Adding to the complexity is the limited availability of publicly accessible experimental data for woven composites, which is crucial for developing comprehensive failure criteria. This thesis presents an experimental and analytical study aimed at providing the next step towards addressing the above discussed shortcomings. An in-depth review of existing failure criteria for fibre-reinforced composites is first presented. This is followed by an extensive experimental test campaign that included both in-plane and through-thickness characterisation tests, investigating the failure of an aerospace indus\-try-grade woven thermoset carbon fibre-reinforced composite material system. The tests were carried out under both tensile and compressive loading conditions, for a variety of on- and off-axis specimen geometries, enabling a better understanding of failure for a number of combined loading scenarios. These tests should serve as a foundational step in addressing the lack of experimental data in the literature to determine a failure criterion for woven composites. A modified version of the Rolfes failure criterion is proposed, which is suitable for orthogonal woven composites. The modified version is based on a reduced set of required stress states and the corresponding failure strengths, allowing for a more general determination of the failure function while requiring fewer experiments. The failure surface of the modified criterion is evaluated against the experimental data obtained. Finally, the performance of the criterion is reviewed, identifying both strengths and weaknesses. The results show that additional improvements are required for the proposed failure criterion. It builds an acceptable basis for the in-plane failure prediction. However, considerable enhancements are required to approximate the behaviour of out-of-plane failures and failure under biaxial stress conditions. These adaptations are essential to enable the application of the criterion in predicting material failure of carbon fibre-reinforced orthogonal woven composites. Furthermore, the limitations of the failure criterion also emphasise the need for further experimental testing, to better populate several of the three-dimensional failure stress states which remain sparsely filled with the tests conducted to date.