By morphing their wings and tail, birds manoeuvre around obstacles and mitigate the effects of atmospheric turbulence with apparent ease. However, the stability and control of bird flight is poorly understood due to difficulty obtaining the relevant data. In this project, linear flight dynamics models of gliding birds were created based on rigid body assumptions and small perturbations from trim. These represent the first flight dynamics models of birds based on shapes and mass properties closely matching those in free-flight. A novel multi-stereo approach was used to reconstruct the surfaces of a free-gliding barn owl (Tyto alba) and peregrine falcon (Falco peregrinus) at a single instant in time during steady gliding flight. The surface reconstructions were used to create vortex lattice models for three flights per bird. These models were integrated with centre of mass and moment of inertia estimates from calibrated X-ray computed tomography scans of barn owl and peregrine cadavers. Linear flight dynamics models based on these datasets revealed a high degree of longitudinal instability in both birds. The time to double of the pitch divergence was typically below 50 ms, which is three times faster than the highly unstable X-29 experimental aircraft. Lateral-directional dynamic stability varied between flights and species, particularly the dutch roll and spiral modes. Current understanding of avian physiology suggests that neural feedback may be too slow to stabilise these animals. This implies a potential role for passive stabilisation through structural compliance, a mechanism that could increase the time available for neural feedback. Overall, this project revealed new insights into the flight stability of gliding birds, largely through the novel application of the imaging methods used. These findings could inspire stabilisation mechanisms for future designs of unmanned air vehicles of similar size.