This paper reports the development of a numerical material model and associated data that successfully captures the behaviour of a CFRP structure subject to direct hypervelocity impact with and without the addition of Kevlar-epoxy layers bonded to the CFRP. Experimental tests reported in this paper have shown that the presence of the Kevlar-epoxy layers not only reduces the extent of delamination and increases the impact energy required for rear surface damage, but also reduces the generation of CFRP fibres released post-impact. The reported simulations have been carried out using the AUTODYN-2D hydrocode software in which it is possible to couple orthotropic constitutive behaviour with a non-linear equation of state. A programme of physical testing was developed to facilitate derivation of the material constants required by the CFRP material model. An extensive series of static tensile and compressive tests was performed in order to obtain the in-plane properties. Interlamina shear stress was also measured. Inverse flyer plate tests were carried out, from which the through thickness dynamic properties were derived. The model is validated against light gas gun tests performed over a range of impact velocities and projectile diameters. A good correspondence with experiment is achieved. Simulation results are also presented for the hybrid structure design with a Kevlar-epoxy layer bonded on either side of the CFRP laminate. Again, a good level of correspondence is seen with corresponding light gas gun tests. Damage observed in the experiments is measured from x-ray images of the impacted targets. The extent of damage within the target is clearly identified in x-ray images taken with iodine penetrant applied to the target. With a high degree of confidence in the numerical models, a sensitivity study is performed to investigate the optimum hybrid structure configuration. The impact hazard will change depending on the spacecraft size, mission lifetime, altitude and inclination and such a modelling approach provides an efficient way to determine the optimum configuration. (C) 2003 Elsevier Ltd. All rights reserved.